Antibodies are critical reagents used in a variety of assays and protocols in biomedical and clinical research. The ability to detect, quantify, enrich, localize, and/or perturb the function of a target protein – even when present in a complex protein mixture, such as a cell lysate or tissue slice, or even in an intact organism – is key to many biomedical research studies. Similarly, the ability to detect changes in protein levels, localization, or interactions with other proteins or membranes, is critical when seeking to identify the pathways involved in cell regulation and disease pathologies.

For these reasons, the market for antibodies has soared, growing from ~10,000 commercially available antibodies about 15 years ago, to more than six million today (Longworth and Chalmers, 2022). However, it has been estimated that ~50% of commercial antibodies fail to meet even basic standards for characterization, and this problem is thought to result in financial losses of $0.4–1.8 billion per year in the United States alone (Ayoubi et al., 2023; Bradbury and Plückthun, 2015; Voskuil et al., 2020; Baker, 2015b). Moreover, the problems caused by the variable quality and characterization of commercial antibodies are compounded by end users not receiving sufficient training in the identification and use of suitable antibodies.

Together these issues have led to an alarming increase in the number of scientific publications that contain misleading or incorrect interpretations and conclusions because they are based on data from experiments that used antibodies that had not been properly characterized or validated (Goodman, 2018; Menke et al., 2020; Laflamme et al., 2019; Aponte Santiago et al., 2023). This situation, and the resulting problems with reproducibility, has been termed a ‘crisis’ (see, for example, Baker, 2015b). There is also a growing body of data that includes stark demonstrations of the volume of incorrect or misleading data published, including clinical patient trials (see, for example, Andersson et al., 2017, Virk et al., 2019, and table 1 in Voskuil et al., 2020), based upon the use of poorly characterized antibodies. (It should be noted, however, that therapeutic antibodies – unlike research antibodies – are very well regulated and are subject to strict controls involving manufacturer and clinical trials; Longworth and Chalmers, 2022).

The roots of this now pervasive problem can be traced back to the early 2000 s, when the first near-complete human genome sequence became available, and attention turned to using this information to make further discoveries in biomedical sciences. Analyses of the genome resulted in open questions as to how many proteins are encoded and which are expressed in different tissues. Such discussions resulted in formation of the Human Proteome Organization (Omenn, 2021) and its Human Proteome Project (Legrain et al., 2011). As suggested by the name, the Human Proteome Project focused on the determination of the human proteome, based upon experimental evidence, and the development of tools (notably mass spectrometry and antibodies) to enable further research. They also developed three foundational approaches to determine the human proteome and make it available to researchers: (i) shotgun and targeted mass spectrometry; (ii) polyclonal and monoclonal antibodies; (iii) an integrated database, intended to promote sharing of data and a platform for multiple uses.

The Human Protein Atlas was launched around the same time and had similar goals: to map all human proteins in cells, tissues, and organs. This project’s approach was highly dependent upon use of antibodies in immunohistochemistry and immunofluorescence (Berglund et al., 2008). Thus, as research efforts turned from the genome to the proteome, there was a clear need for a large increase in the number of antibodies with the requisite affinities and specificities to meet the needs of these initiatives. However, over the following 20 years, concern over the lack of consensus for how best to validate antibody usage grew and, perhaps worse, it became clear that many end users did not adequately understand how the quality of their data depended on the antibodies they used being properly validated (see, for example, Gloriam et al., 2010; Bandrowski, 2022; Singh Chawla, 2015; Blow, 2017; Lund-Johansen, 2023; Couchman, 2009; Williams, 2018; Pillai-Kastoori et al., 2020, Andersson et al., 2017; Sivertsson et al., 2020; Polakiewicz, 2015; Howat et al., 2014; Edfors et al., 2018; Gilda et al., 2015).

In the early days of antibody research most antibodies were generated and characterized in research labs, and sent to other research groups upon request. But as demand increased, companies selling antibodies to researchers became increasingly important. At first these companies relied on researchers supplying them with antibodies that had already been characterized, and the researchers received some fraction of the sale price in return.

Later, many companies started to generate the antibodies themselves, particularly for the most widely used antibodies, as these were the most profitable. It is important for anyone hoping to understand and address the antibody crisis to realise that the costs associated with performing even the most basic/common characterizations (Western blotting, immunoprecipitation, immunofluorescence and immunohistochemistry), and obtaining appropriate knockout cell lines, is many times higher than the revenues generated from the average antibody on the market today. Thus, the current system puts the onus on end users to both find the best antibody on the market for them, and to perform appropriate characterization prior to using the antibody – otherwise they could waste a lot of time and money on experiments that do not produce meaningful or trustworthy results. Finding the best antibody for your research, and then validating it, can be quite challenging, so below we describe some resources that can help.

A related issue is that with the focus placed on the most widely used antibodies, the research community is failing to take advantage of all the information contained in the human genome sequence because the high-quality reagents (including antibodies) needed to investigate a large fraction of the proteome are simply not available or identifiable as a result of the lack of characterization data linked to them (Edwards et al., 2011).

The origins of this antibody crisis are many and varied, so it is unrealistic to expect a complete ‘solution’ in the near future. However, there are a number of steps can be taken to dramatically improve the situation, such as: continuing to raise awareness of the issues among end users; more accurately identifying the reagents on vendor websites and in scientific publications; and sharing characterization data, when available.

In this article we assemble a history of key events in the field (Figure 1); summarize efforts to raise awareness of the problem – and ways to address the problem – through editorials, meetings, and workshops; and discuss a number of initiatives that are trying to address the problem. The role of technology development in the field, and its role in planning future initiatives, is also discussed. In the last section we propose actions for the various stakeholders – researchers and end users, universities, journals, antibody vendors and repositories, scientific societies, charities and funding agencies – that we believe will help address the antibody characterization and its huge, ongoing and detrimental impact on reproducibility. Clearly, a consensus amongst all stakeholders is required to resolve the antibody characterization crisis.

Figure 1

Download asset Open asset

The term antibody ‘characterization’ is more apt than ‘validation’, as an antibody’s suitability often varies across different assays. The term antibody ‘characterization’ is best applied to a description of the inherent ability of an antibody to perform in different assays – i.e. for a recombinant antibody it describes the properties of an antibody with a specific sequence (e.g. functional in Western blot, immunohistochemistry and immunofluorescence, but not in immunoprecipitation). While validation is best applied to confirmation that a particular antibody lot received in a lab performs as characterized – an antibody may not be validated if (for example) it has been inadvertently denatured by environmental extremes.

Antibody characterization can be thought of as simply the controls needed in any protocol that uses an antibody, to ensure that the antibody is performing with the requisite specificity needed to ensure accurate interpretation of the results. In contrast, the term validation is more appropriate when demonstrating context-specific suitability for an application. Also, the terms ‘specificity’ and ‘selectivity’ are often used interchangeably, though the former refers to the ability to bind only to its intended target, while the latter refers to its ability to bind that protein even when in a complex mixture of proteins. For a concise introduction to the use of antibodies in research, particularly in immunohistochemistry, including the basic principles of antibody action and the different types of antibodies, see Saper, 2009.

Readers should also be aware that some terms in the field are used interchangeably, even though they mean slightly different things, which can be confusing. For example, immunofluorescence, immunohistochemistry, and immunocytochemistry are all used to localize targets in cells or tissues, which may involve the use of a variety of different fixatives or permeabilization steps. However, the key differences are in the method of detection: the term immunofluorescence should be used when fluorescence is the output, while the other terms are more appropriate when the readout is generated by enzymatic labels, such as peroxidase and alkaline phosphatase. Also, immunofluorescence can be used on live cells, while the other two approaches require fixation. In this article we will discuss just immunofluorescence and immunohistochemistry for the sake of simplicity.

Complexity arises from the use of antibodies across different experimental protocols, each requiring distinct controls. Common applications include Western blotting (also known as immunoblotting), immunoprecipitation, immunofluorescence on cultured cells, immunohistochemistry on tissue sections, flow cytometry with intact cells in suspension, enzyme-linked immunosorbent assay (ELISA), and functional assays. Characterization of an antibody should include testing in as many of these assays as feasible to determine its potential uses and value to researchers. However, the combination of studies needed to validate an antibody for a particular use depends upon the experiment that is being undertaken. In other words, characterization data from vendors, publications and public databases can be helpful when identifying the best candidate antibodies for a specific use, but researchers should always confirm that the antibodies they plan to use will perform as needed in their experiments.

The prevalent use of polyclonal antibodies, derived from immunized animals and most often used in the form of serum from such animals, is a large source of the problems described herein. This is due to their non-renewable nature and the complexity of the different antibodies present, which can influence batch variability as a result of the presence of both specific and non-specific antibodies. This may introduce false positives and increased background noise, and is further confounded when serum from different bleeds or animals is sold under the same name and catalog number. Also, the profile of a polyclonal antibody response can vary over time, even with affinity purification, as the antibody population in each batch is varied. Despite these drawbacks, polyclonal antibodies remain widely used, necessitating a coordinated characterization effort.

Monoclonal antibodies have historically been generated by fusion of an immortal myeloma cell line with terminally differentiated B cells from an immunized animal. Cloning of the resulting hybridoma cells, that each secrete a single antibody, allows one to identify antibodies that bind the target protein in different assays and to generate large batches of that single antibody. Thus, monoclonal antibodies were the first type of renewable antibody as they could be generated repeatedly and would remain consistent. However, caution is advised as hybridoma lines can vary over time and may express more than one antibody (Bradbury et al., 2018; Mitchell et al., 2023).

Recombinant antibodies represent a newer technology, involving the determination of the DNA sequence encoding the antigen binding site. This allows cloning it into plasmids to allow expression of a single antibody, offering stable and renewable reagents, with customizable constant regions for varied applications and much more flexibility in use. Newer technologies allow culture of single B cells from an immunized animal, allowing cloning of the secreted antibody to generate a recombinant antibody. The use of large libraries that encode many millions of potential antigen binding regions, displayed on the surface of phage particles or yeast cells, has allowed high-throughput screening for high affinity reagents that have the added value of not requiring the use of any animals in their development. Some researchers believe that an animal’s immune system may still do a better job of identifying the best epitopes for targeting than even a large library of phage or yeast cells, though this remains somewhat controversial (Custers and Steyaert, 2020).

Herein, we focus on antibodies that target proteins (as opposed to polysaccharides, nucleic acids or other cellular components) and use the term epitope to indicate the specific sites/residues within a given target protein (antigen) to which an antibody binds. It is worth noting here that the same epitope can be present on other proteins. The ideal antibody would bind to a unique, linear epitope that is fully exposed on the surface of the folded protein and not involved in any post-translational modifications or protein–protein interactions, allowing equal antibody access in both native and unfolded conditions and without regard to tissue/cell-specific differences in protein associations or post-translational modifications. In contrast, there are many examples of very useful antibodies that bind the antigen in a conformation dependent fashion, and thus fail to bind to the denatured protein (in, for example, Western blotting). Similarly, antibodies that bind to epitopes that include specific post-translational modifications can be valuable for monitoring such modifications over time and space. While potentially valuable reagents, such antibodies present challenges in characterization that are not discussed here. The affinity of any antibody should be low nanomolar or tighter, to allow detection of low abundance, endogenous protein, while maintaining high specificity for a single protein.

In the early 2000s, researchers and journals began raising the alarm about the growing antibody crisis, explaining why appropriate controls and detailed/accurate reporting of antibodies was so important (Saper, 2005; Saper, 2009; Saper and Sawchenko, 2003). In a review article titled ‘Antibody Validation’ that was published in 2010, Bordeaux and co-authors discussed the importance of antibody characterization in some detail (Bordeaux et al., 2010), while also stressing the need to differentiate between the lack of antibody characterization and scientific misconduct or data manipulation (Collins and Tabak, 2014). This article also underscored common pitfalls in antibody use, including a lack of appropriate user training, over-reliance on vendors for characterization, inadequate methodological details from providers and in publications, and challenges in reagent identification. Bordeaux et al. also highlighted the problems that emerge when antibodies are used without characterization, and cited numerous examples of such misuse.

While we agree with the statement that ‘the responsibility for proof of specificity is with the purchaser, not the vendor’ (Bordeaux et al., 2010), we argue herein that all stakeholders have responsibilities when it comes to addressing the antibody crisis. Moreover, we encourage the use of knockout (KO) or knockdown (KD) cell lines or tissue samples as important negative controls for specificity. KO cell lines and model organisms have become much more readily available, thanks to CRISPR technologies, and the use of antibodies to confirm KO of a protein is just as useful a tool as the use of the KO lines to test for specificity of antibodies. Unfortunately, there is currently no repository that researchers can use to share KO cell lines. As useful as KO cells/tissues are as negative controls for specificity, the characterization of antibodies is always further improved when combined with other approaches.

A number of articles in various Nature journals contributed to the conversation about the antibody crisis, including an editorial (Methods, 2015), various news articles (Baker, 2015b; Baker, 2015a; Baker, 2016; Baker, 2020), a comment article about reproducibility issues in preclinical cancer research (Begley and Ellis, 2012), and another comment article with over 100 co-signatories that asked the National Institutes of Health and European Union “to convene academic users, technology developers, biotech companies, funding agencies, and publishers, and establish a realistic timetable for the transition to these [recombinant antibodies] high quality binding reagents” (Bradbury and Plückthun, 2015).

The Global Biological Standards Institute (GBSI), established in 2012, played a significant role by forming a Research Antibodies and Standards Task Force, and its 2015 online survey and subsequent analyses (Freedman et al., 2016) emphasized the urgent need for better training for end users and for standards for best practices. Another GBSI study found that the US spends ~$28 billion per year on preclinical research that is not reproducible, and concluded that urgent improvements were required in two areas – study design and the validation of biological reagents (Freedman et al., 2015). The GBSI has also emphasized the need for the validation of cell lines, not just antibodies (Souren et al., 2022; Marx, 2014). A GBSI meeting in Asilomar in 2016 focused on developing antibody characterization standards (Baker, 2016), and although the scoring system proposed at this meeting has seen limited use, webinars developed in concert with the Antibody Society Voskuil et al., 2020 have proven valuable in educating users.

The Federation of American Societies of Experimental Biology (FASEB) also published an important report called Enhancing Research Reproducibility in 2016, which stressed the need for standard reporting formats for antibodies (FASEB, 2016). More recently, in 2023, the annual meeting of the American Society for Cell Biology included a workshop on antibody characterization. One conclusion from this workshop was that venders and researchers “need to adopt a standard format for reporting antibody information.”

There has also been a series of Alpbach Workshops on Affinity Proteomics that have discussed aspects of antibody generation and characterization. A write up of their 2017 meeting summarized discussions around topics that included antibody specificity being ‘context-dependent’ and characterization needing to be performed by end users for each specific use (Taussig et al., 2018). They also emphasized the fact that characterization data are potentially cell or tissue type specific. They included a summary of efforts underway at the time to establish characterization guidelines, and the roles that publishers and authors can and should play in addressing the optimal use and reporting of antibody-based experiments. At the most recent workshop (March 2024), representatives from various companies presented recombinant antibody or binder generation technologies. Participants endorsed these, particularly after demonstrations by representatives from YCharOS and Abcam using KO cell lines, which showed that recombinant antibodies were more effective than polyclonal antibodies, and far more reproducible.

An ad hoc International Working Group for Antibody Validation was formed in 2016 with a goal of addressing the “collective need for standards to validate antibody specificity and reproducibility, as well as the need for reporting practices” (Uhlen et al., 2016). This group introduced the ‘five pillars’ of antibody characterization (Table 1): (i) genetic strategies (i.e., the use of knockout and knockdown techniques as controls for specificity); (ii) orthogonal strategies (i.e., comparing the results of antibody-dependent and antibody-independent experiments); (iii) multiple (independent) antibody strategies (which compare the results of experiments that use different antibodies to target the same protein); (iv) recombinant strategies (which increase target protein expression); (v) immunocapture MS strategies (in which mass spectrometry is used to identify the protein(s) captured by the antibody). These pillars are not intended to encompass all useful characterization strategies, nor are they all required for each characterization effort: rather, users are encouraged to use as many as feasible (Uhlen et al., 2016).

In summary, in order to generate reliable data when using antibodies in an experiment, the characterization of the antibody needs to document the following: (i) that the antibody is binding to the target protein; (ii) that the antibody binds to the target protein when in a complex mixture of proteins (e.g., whole cell lysate or tissue section); (iii) that the antibody does not bind to proteins other than the target protein; (d) that the antibody performs as expected in the experimental conditions used in the specific assay employed. These articles and workshops certainly helped raise awareness of the issues, identified each of the key concerns, and laid out some strategies for improvement.

Numerous international efforts have been initiated to address challenges in antibody characterization, particularly targeting the human proteome. It is worth noting the incredible value of proteome-wide approaches, not just for studies of human proteins, but for all model organisms used in basic and biomedical research. What is learned from targeting the human proteome will benefit efforts targeting other proteomes, and vice versa.

The early, large-scale efforts typically focused on the use of high-throughput screening and assays, as well as the use of non-antibody binding molecules (such as protein affinity reagents; Gloriam et al., 2010; Taussig et al., 2007). While these early projects generated some useful reagents and data, they fell short of their initial goals, although they did help to reveal the scale of the challenges and the limitations of the approaches being used. Below is a summary of these efforts plus two related efforts – the Research Resource Identifier (RRID) program, and the Developmental Studies Hybridoma Bank (DSHB) – presented in a somewhat chronological order (Figure 1).

Before embarking on large, publicly-funded projects, a crucial question is whether to wait for new technologies that might expedite success, enhance output quality, or reduce costs. This dilemma was evident in the human genome sequencing discussions, where starting the project proved valuable despite initial technological limitations. Note that there is a clear difference between the work involved in generating optimal antibody reagents and in characterizing them. This prompts the question: Is there a need to wait for newer technologies that will expedite and improve the processes involved in antibody characterization?

The primary challenge in characterizing antibodies lies in the diversity of assays for antibody usage, the variations within each assay, and the incompatibility of common assays with high-throughput methods. To establish realistic goals, one needs a finite set of assays, with standardized protocols agreed upon by stakeholders. These characterization efforts should be performed using assays similar to those used by end users.

Towards this end, members of the YCharOS team and representatives from ten leading antibody manufacturers recently co-wrote a method article that contains detailed protocols for Western Blots, immunoprecipitation and immunofluorescence (Ayoubi et al., 2024). A consensus on such assays – made possible by collaborations among stakeholders – could lead to a pipeline identifying reagents that are effective for these assays. Most antibodies will work in some assays but not all. Thus, even if an antibody does not perform well in some characterization trials, it may be valuable in other assays or under other conditions than those used previously. Thus, all data used in these evaluations, and the protocols used, should be openly shared to allow end users to assess the data as they deem most appropriate to their needs. Even with such a pipeline in place, end users must still validate the antibodies they plan to use in their own work.

A challenge in large-scale antibody characterization is the current lack of high-throughput assays (including ones that might be readily automated) that are similar enough to end user assays to be optimal for characterization. While peptide or fragment display approaches exist, extrapolating their data to more common assays is risky. However, a combination of immunoprecipitation and mass spectrometry can be used to speed and improve the characterization of the selectivity and specificity of antibodies at scale in this one use (Marcon et al., 2015).

In contrast to the limited role that high-throughput assays are likely to play in the characterization of antibodies, technologies allowing rapid screening of large libraries for high-affinity, specific recombinant antibodies are generating large numbers of new antibodies, some with affinities better than those obtainable by immunization (Azevedo Reis Teixeira et al., 2021). These recombinant antibodies will still require characterization in all intended assays, but once characterized will significantly ameliorate problems related to lot to lot variation. We can expect the number of commercially available recombinant antibodies to continue to grow rapidly in the coming years, only further arguing for the need to gain consensus on issues surrounding their characterization, proper use, and reporting.

Finally, a role for technology in this field that will likely play increasingly important roles in the future development and optimization of antibodies will come from deep learning models, such as AlphaFold. These models should enable predictions of the antibody-antigen complex, aid in the identification of the epitope targeted, and help determine if folding, post-translational modifications or other issues may influence the output from use of the antibody. These future prospects are another reason to encourage open access to the antibody sequences of recombinant antibodies.

For many years end users have been angered by the waste of time and money associated with bad antibodies, and have been frustrated by the inability to reproduce work from other labs.

The goals of providing researchers with optimal reagents and enhancing the reproducibility of antibody-based data depends on all the relevant stakeholders understanding the issues involved and working collaboratively towards improvement. At a minimum this involves: (i) development of consensus in characterization standards; (ii) the use of open source(s) for depositing characterization data, thereby ensuring access and reducing redundancy; (iii) rigorous use of reporting tools to clearly and unambiguously identify reagents used (such as RRID and SciScore); (iv) prioritized engagement with active participants to incentivize and speed the adoption of best practices for reproducibility; and (v) the widespread adoption of recombinant antibodies.

We list below the various stakeholders, along with recommendations for what they can do to improve the current situation. In an ideal long-term scenario, only recombinant antibodies should be used, and they should be characterized for specificity by testing on KO cell lines in the specific assays in which they will used. While it would be ideal to have antibody sequences publicly available, this remains unlikely for commercially sourced ones. That said, it appears useful to explore concepts around providing RRID-like identifiers to recombinant antibodies known to have identical sequences, allowing researchers to at least know they are using antibodies of identical sequence, without knowing the sequences themselves.

Over the past two decades, many individuals and projects have contributed significantly to addressing the crisis stemming from inadequate characterization of antibodies used in research. Efforts to generate and make available antibodies for the entire human proteome have illuminated the complexities of the issues, emphasizing the need for a multi-faceted approach and the involvement of all stakeholders. Key achievements include bringing antibody characterization to the forefront of stakeholder concerns, and separating the objectives of antibody generation from characterization. Newer technologies promise to accelerate the creation of new recombinant antibodies (so researchers would no longer have to use polyclonal antibodies). However, the characterization of these new reagents – and the six million or so antibodies currently on the market – remains crucial, necessitating data sharing, identification of optimal reagents, and the removal of non-specific antibodies. Let us be clear, there is no one ‘fix’ to the problems outlined here. The problems are ongoing and will continue. Raising awareness of the issues and holding all stakeholders accountable for contributing to improvements are viewed as the best approaches to minimizing future waste and damage.

Each stakeholder has a distinct motivation yet a shared commitment to supporting well-controlled antibody-based studies. Researchers strive for high-quality data to drive novel discoveries, quality publications, and paths towards improving human health. Universities aim to support research and provide comprehensive training. Disease foundations focus on identifying key reagents for disease pathways, while reputable journals and publishers aim to uphold high scientific standards. Vendors have the responsibility to provide characterized products, a goal some are now pursuing through collaborations with groups like YCharOS and use of RRIDs. Funding agencies have supported initiatives in the past and should continue to promote high-quality, reproducible research, taking advantage of what has been learned from prior projects and building upon it to improve the quality of the reagents and the resulting data.

The community is encouraged to support and further develop pipelines like the YCharOS pipeline, which exemplify collaborative efforts towards consensus on characterization assays and public data sharing. The use of tools like RRID and SciScore should become standard practice, easing the burden of enforcing necessary changes and enhancing the overall quality and reproducibility of antibody-based research. Scientists can build consortia in their areas of expertise to share the costs of characterization and make readily available the best possible reagents with agreed upon characterization testing and uses.

No data were generated for this work.

1. 1. Acharya P
   2. Quinlan A
   3. Neumeister V

   (2017) The ABCs of finding a good antibody: How to find a good antibody, validate it, and publish meaningful data

   F1000Research 6:851.

   https://doi.org/10.12688/f1000research.11774.1

   • PubMed
   • Google Scholar
2. 1. Andersson S
   2. Sundberg M
   3. Pristovsek N
   4. Ibrahim A
   5. Jonsson P
   6. Katona B
   7. Clausson CM
   8. Zieba A
   9. Ramström M
   10. Söderberg O
   11. Williams C
   12. Asplund A

   (2017) Insufficient antibody validation challenges oestrogen receptor beta research

   Nature Communications 8:15840.

   https://doi.org/10.1038/ncomms15840

   • PubMed
   • Google Scholar
3. 1. Andrews NP
   2. Boeckman JX
   3. Manning CF
   4. Nguyen JT
   5. Bechtold H
   6. Dumitras C
   7. Gong B
   8. Nguyen K
   9. van der List D
   10. Murray KD
   11. Engebrecht J
   12. Trimmer JS

   (2019) A toolbox of IgG subclass-switched recombinant monoclonal antibodies for enhanced multiplex immunolabeling of brain

   eLife 8:e43322.

   https://doi.org/10.7554/eLife.43322

   • PubMed
   • Google Scholar
4. 1. Aponte Santiago N
   2. Konecki S
   3. Porter L
   4. Virk H

   (2023) Tales of the unexpected

   eLife 12:e87444.

   https://doi.org/10.7554/eLife.87444

   • PubMed
   • Google Scholar
5. 1. Ayoubi R
   2. Ryan J
   3. Biddle MS
   4. Alshafie W
   5. Fotouhi M
   6. Bolivar SG
   7. Ruiz Moleon V
   8. Eckmann P
   9. Worrall D
   10. McDowell I
   11. Southern K
   12. Reintsch W
   13. Durcan TM
   14. Brown C
   15. Bandrowski A
   16. Virk H
   17. Edwards AM
   18. McPherson P
   19. Laflamme C

   (2023) Scaling of an antibody validation procedure enables quantification of antibody performance in major research applications

   eLife 12:RP91645.

   https://doi.org/10.7554/eLife.91645

   • PubMed
   • Google Scholar
6. 1. Ayoubi R
   2. Ryan J
   3. Bolivar SG
   4. Alende C
   5. Moleon VR
   6. Fotouhi M
   7. Southern K
   8. Alshafie W
   9. Baker MR
   10. Ball AR
   11. Callahan D
   12. Cooper JA
   13. Crosby K
   14. Harvey KJ
   15. Houston DW
   16. Kumaran R
   17. Rego M
   18. Schofield C
   19. Wu H
   20. Biddle MS
   21. Brown CM
   22. Kahn RA
   23. Bandrowski A
   24. Virk HS
   25. Edwards AM
   26. McPherson PS
   27. Laflamme C

   (2024) A consensus platform for antibody characterization

   Protocol Exchange.

   https://doi.org/10.21203/rs.3.pex-2607/v1

   • Google Scholar
7. 1. Azevedo Reis Teixeira A
   2. Erasmus MF
   3. D’Angelo S
   4. Naranjo L
   5. Ferrara F
   6. Leal-Lopes C
   7. Durrant O
   8. Galmiche C
   9. Morelli A
   10. Scott-Tucker A
   11. Bradbury ARM

   (2021) Drug-like antibodies with high affinity, diversity and developability directly from next-generation antibody libraries

   mAbs 13:1980942.

   https://doi.org/10.1080/19420862.2021.1980942

   • PubMed
   • Google Scholar
8. 1. Baker M

   (2015a) Antibody anarchy: A call to order

   Nature 527:545–551.

   https://doi.org/10.1038/527545a

   • PubMed
   • Google Scholar
9. 1. Baker M

   (2015b) Reproducibility crisis: Blame it on the antibodies

   Nature 521:274–276.

   https://doi.org/10.1038/521274a

   • PubMed
   • Google Scholar
10. 1. Baker M

    (2016) Biologists plan scoring system for antibodies

    Nature.

    https://doi.org/10.1038/nature.2016.20712

    • Google Scholar
11. 1. Baker M

    (2020) When antibodies mislead: The quest for validation

    Nature 585:313–314.

    https://doi.org/10.1038/d41586-020-02549-1

    • PubMed
    • Google Scholar
12. 1. Bandrowski A
    2. Brush M
    3. Grethe JS
    4. Haendel MA
    5. Kennedy DN
    6. Hill S
    7. Hof PR
    8. Martone ME
    9. Pols M
    10. Tan SC
    11. Washington N
    12. Zudilova-Seinstra E
    13. Vasilevsky N

    (2016) The Resource Identification Initiative: A cultural shift in publishing

    Journal of Comparative Neurology 524:8–22.

    https://doi.org/10.1002/cne.23913

    • PubMed
    • Google Scholar
13. 1. Bandrowski A

    (2022) A decade of GigaScience: What can be learned from half a million RRIDs in the scientific literature?

    GigaScience 11:giac058.

    https://doi.org/10.1093/gigascience/giac058

    • PubMed
    • Google Scholar
14. 1. Bandrowski A
    2. Pairish M
    3. Eckmann P
    4. Grethe J
    5. Martone ME

    (2023) The Antibody Registry: Ten years of registering antibodies

    Nucleic Acids Research 51:D358–D367.

    https://doi.org/10.1093/nar/gkac927

    • PubMed
    • Google Scholar
15. 1. Begley CG
    2. Ellis LM

    (2012) Drug development: Raise standards for preclinical cancer research

    Nature 483:531–533.

    https://doi.org/10.1038/483531a

    • PubMed
    • Google Scholar
16. 1. Berglund L
    2. Björling E
    3. Oksvold P
    4. Fagerberg L
    5. Asplund A
    6. Al-Khalili Szigyarto C
    7. Persson A
    8. Ottosson J
    9. Wernérus H
    10. Nilsson P
    11. Lundberg E
    12. Sivertsson Å
    13. Navani S
    14. Wester K
    15. Kampf C
    16. Hober S
    17. Pontén F
    18. Uhlén M

    (2008) A genecentric human protein atlas for expression profiles based on antibodies

    Molecular & Cellular Proteomics 7:2019–2027.

    https://doi.org/10.1074/mcp.R800013-MCP200

    • Google Scholar
17. 1. Biddle M
    2. Stylianou P
    3. Rekas M
    4. Wright A
    5. Sousa J
    6. Ruddy D
    7. Stefana MI
    8. Kmiecik K
    9. Bandrowski A
    10. Kahn RA
    11. Laflamme C
    12. Krockow EM
    13. Virk HS

    (2024) Improving the integrity and reproducibility of research that uses antibodies: A technical, data sharing, behavioral and policy challenge

    mAbs 16:2323706.

    https://doi.org/10.1080/19420862.2024.2323706

    • PubMed
    • Google Scholar
18. 1. Björling E
    2. Uhlén M

    (2008) Antibodypedia, a portal for sharing antibody and antigen validation data

    Molecular & Cellular Proteomics 7:2028–2037.

    https://doi.org/10.1074/mcp.M800264-MCP200

    • PubMed
    • Google Scholar
19. 1. Blackshaw S
    2. Venkataraman A
    3. Irizarry J
    4. Yang K
    5. Anderson S
    6. Campbell E
    7. Gatlin CL
    8. Freeman NL
    9. Basavappa R
    10. Stewart R
    11. Loss MA
    12. Pino I
    13. Zhu H
    14. Bader JS

    (2016) The NIH Protein Capture Reagents Program (PCRP): A standardized protein affinity reagent toolbox

    Nature Methods 13:805–806.

    https://doi.org/10.1038/nmeth.4013

    • PubMed
    • Google Scholar
20. 1. Blow N

    (2017) Fighting for the future of antibodies

    BioTechniques 63:51–56.

    https://doi.org/10.2144/000114573

    • PubMed
    • Google Scholar
21. 1. Bordeaux J
    2. Welsh A
    3. Agarwal S
    4. Killiam E
    5. Baquero M
    6. Hanna J
    7. Anagnostou V
    8. Rimm D

    (2010) Antibody validation

    BioTechniques 48:197–209.

    https://doi.org/10.2144/000113382

    • PubMed
    • Google Scholar
22. 1. Bradbury A
    2. Plückthun A

    (2015) Reproducibility: Standardize antibodies used in research

    Nature 518:27–29.

    https://doi.org/10.1038/518027a

    • PubMed
    • Google Scholar
23. 1. Bradbury ARM
    2. Trinklein ND
    3. Thie H
    4. Wilkinson IC
    5. Tandon AK
    6. Anderson S
    7. Bladen CL
    8. Jones B
    9. Aldred SF
    10. Bestagno M
    11. Burrone O
    12. Maynard J
    13. Ferrara F
    14. Trimmer JS
    15. Görnemann J
    16. Glanville J
    17. Wolf P
    18. Frenzel A
    19. Wong J
    20. Koh XY
    21. Eng H-Y
    22. Lane D
    23. Lefranc M-P
    24. Clark M
    25. Dübel S

    (2018) When monoclonal antibodies are not monospecific: Hybridomas frequently express additional functional variable regions

    mAbs 10:539–546.

    https://doi.org/10.1080/19420862.2018.1445456

    • PubMed
    • Google Scholar
24. Website

    1. CiteAb

    (2024) Finding the right antiobody

    Accessed July 23, 2024.

    https://www.citeab.com/finding-the-right-antibody
25. 1. Collins FS
    2. Tabak LA

    (2014) Policy: NIH plans to enhance reproducibility

    Nature 505:612–613.

    https://doi.org/10.1038/505612a

    • PubMed
    • Google Scholar
26. 1. Couchman JR

    (2009) Commercial antibodies: The good, bad, and really ugly

    Journal of Histochemistry and Cytochemistry 57:7–8.

    https://doi.org/10.1369/jhc.2008.952820

    • PubMed
    • Google Scholar
27. 1. Custers R
    2. Steyaert J

    (2020) Discussions on the quality of antibodies are no reason to ban animal immunization

    EMBO Reports 21:e51761.

    https://doi.org/10.15252/embr.202051761

    • PubMed
    • Google Scholar
28. 1. Davies P
    2. Hinkle KM
    3. Sukar NN
    4. Sepulveda B
    5. Mesias R
    6. Serrano G
    7. Alessi DR
    8. Beach TG
    9. Benson DL
    10. White CL
    11. Cowell RM
    12. Das SS
    13. West AB
    14. Melrose HL

    (2013) Comprehensive characterization and optimization of anti-LRRK2 (leucine-rich repeat kinase 2) monoclonal antibodies

    Biochemical Journal 453:101–113.

    https://doi.org/10.1042/BJ20121742

    • PubMed
    • Google Scholar
29. 1. Edfors F
    2. Hober A
    3. Linderbäck K
    4. Maddalo G
    5. Azimi A
    6. Sivertsson Å
    7. Tegel H
    8. Hober S
    9. Szigyarto CAK
    10. Fagerberg L
    11. von Feilitzen K
    12. Oksvold P
    13. Lindskog C
    14. Forsström B
    15. Uhlen M

    (2018) Enhanced validation of antibodies for research applications

    Nature Communications 9:4130.

    https://doi.org/10.1038/s41467-018-06642-y

    • PubMed
    • Google Scholar
30. 1. Edwards AM
    2. Isserlin R
    3. Bader GD
    4. Frye SV
    5. Willson TM
    6. Yu FH

    (2011) Too many roads not taken

    Nature 470:163–165.

    https://doi.org/10.1038/470163a

    • PubMed
    • Google Scholar
31. Website

    1. FASEB

    (2016) Enhancing Research Reproducibility

    Accessed July 18, 2024.

    https://www.faseb.org/FASEB/media/PDF/News/Washington%20Update/FASEB_Enhancing-Research-Reproducibility_1.pdf
32. 1. Freedman LP
    2. Cockburn IM
    3. Simcoe TS

    (2015) The economics of reproducibility in preclinical research

    PLOS Biology 13:e1002165.

    https://doi.org/10.1371/journal.pbio.1002165

    • PubMed
    • Google Scholar
33. 1. Freedman LP
    2. Gibson MC
    3. Bradbury ARM
    4. Buchberg AM
    5. Davis D
    6. Dolled-Filhart MP
    7. Lund-Johansen F
    8. Rimm DL

    (2016) [Letter to the Editor] The need for improved education and training in research antibody usage and validation practices

    BioTechniques 61:16–18.

    https://doi.org/10.2144/000114431

    • PubMed
    • Google Scholar
34. 1. Gilda JE
    2. Ghosh R
    3. Cheah JX
    4. West TM
    5. Bodine SC
    6. Gomes AV

    (2015) Western blotting inaccuracies with unverified antibodies: Need for a Western Blotting Minimal Reporting Standard (WBMRS)

    PLOS ONE 10:e0135392.

    https://doi.org/10.1371/journal.pone.0135392

    • PubMed
    • Google Scholar
35. 1. Gloriam DE
    2. Orchard S
    3. Bertinetti D
    4. Björling E
    5. Bongcam-Rudloff E
    6. Borrebaeck CAK
    7. Bourbeillon J
    8. Bradbury ARM
    9. de Daruvar A
    10. Dübel S
    11. Frank R
    12. Gibson TJ
    13. Gold L
    14. Haslam N
    15. Herberg FW
    16. Hiltke T
    17. Hoheisel JD
    18. Kerrien S
    19. Koegl M
    20. Konthur Z
    21. Korn B
    22. Landegren U
    23. Montecchi-Palazzi L
    24. Palcy S
    25. Rodriguez H
    26. Schweinsberg S
    27. Sievert V
    28. Stoevesandt O
    29. Taussig MJ
    30. Ueffing M
    31. Uhlén M
    32. van der Maarel S
    33. Wingren C
    34. Woollard P
    35. Sherman DJ
    36. Hermjakob H

    (2010) A community standard format for the representation of protein affinity reagents

    Molecular & Cellular Proteomics 9:1–10.

    https://doi.org/10.1074/mcp.M900185-MCP200

    • PubMed
    • Google Scholar
36. 1. Gong B
    2. Murray KD
    3. Trimmer JS

    (2016) Developing high-quality mouse monoclonal antibodies for neuroscience research - Approaches, perspectives and opportunities

    New Biotechnology 33:551–564.

    https://doi.org/10.1016/j.nbt.2015.11.007

    • PubMed
    • Google Scholar
37. 1. Goodman SL

    (2018) The path to VICTORy - A beginner’s guide to success using commercial research antibodies

    Journal of Cell Science 131:jcs216416.

    https://doi.org/10.1242/jcs.216416

    • PubMed
    • Google Scholar
38. 1. Howat WJ
    2. Lewis A
    3. Jones P
    4. Kampf C
    5. Pontén F
    6. van der Loos CM
    7. Gray N
    8. Womack C
    9. Warford A

    (2014) Antibody validation of immunohistochemistry for biomarker discovery: Recommendations of a consortium of academic and pharmaceutical based histopathology researchers

    Methods 70:34–38.

    https://doi.org/10.1016/j.ymeth.2014.01.018

    • PubMed
    • Google Scholar
39. Website

    1. Human Protein Atlas

    (2020) The Human Protein Atlas: A 20-year journey into to the body

    Accessed July 24, 2024.

    https://www.proteinatlas.org/download/HPA_-_a_20-year_journey_into_the_body.pdf
40. 1. Laflamme C
    2. McKeever PM
    3. Kumar R
    4. Schwartz J
    5. Kolahdouzan M
    6. Chen CX
    7. You Z
    8. Benaliouad F
    9. Gileadi O
    10. McBride HM
    11. Durcan TM
    12. Edwards AM
    13. Healy LM
    14. Robertson J
    15. McPherson PS

    (2019) Implementation of an antibody characterization procedure and application to the major ALS/FTD disease gene C9ORF72

    eLife 8:e48363.

    https://doi.org/10.7554/eLife.48363

    • PubMed
    • Google Scholar
41. 1. Lai WKM
    2. Mariani L
    3. Rothschild G
    4. Smith ER
    5. Venters BJ
    6. Blanda TR
    7. Kuntala PK
    8. Bocklund K
    9. Mairose J
    10. Dweikat SN
    11. Mistretta K
    12. Rossi MJ
    13. James D
    14. Anderson JT
    15. Phanor SK
    16. Zhang W
    17. Zhao Z
    18. Shah AP
    19. Novitzky K
    20. McAnarney E
    21. Keogh MC
    22. Shilatifard A
    23. Basu U
    24. Bulyk ML
    25. Pugh BF

    (2021) A ChIP-exo screen of 887 Protein Capture Reagents Program transcription factor antibodies in human cells

    Genome Research 31:1663–1679.

    https://doi.org/10.1101/gr.275472.121

    • PubMed
    • Google Scholar
42. 1. Landegren U

    (2016) AFFINOMICS and the prospects for large-scale protein analyses

    New Biotechnology 33:491–493.

    https://doi.org/10.1016/j.nbt.2015.09.006

    • PubMed
    • Google Scholar
43. 1. Legrain P
    2. Aebersold R
    3. Archakov A
    4. Bairoch A
    5. Bala K
    6. Beretta L
    7. Bergeron J
    8. Borchers C
    9. Corthals GL
    10. Costello CE
    11. Deutsch EW
    12. Domon B
    13. Hancock W
    14. He F
    15. Hochstrasser D
    16. Marko-Varga G
    17. Salekdeh GH
    18. Sechi S
    19. Snyder M
    20. Srivastava S
    21. Uhlen M
    22. Hu CH
    23. Yamamoto T
    24. Paik YK
    25. Omenn GS

    (2011) The Human Proteome Project: Current state and future direction

    Molecular & Cellular Proteomics.

    https://doi.org/10.1074/mcp.M111.009993

    • PubMed
    • Google Scholar
44. 1. Longworth S
    2. Chalmers A

    (2022) CiteAb for researchers and suppliers: How identifying product citations from publications can help accelerate science

    Information Services & Use 42:319–324.

    https://doi.org/10.3233/ISU-220158

    • Google Scholar
45. Website

    1. Longworth S

    (2024) YCharOS antibody characterization data is now linked from the CiteAb antibody search engine

    Accessed July 23, 2024.

    https://blog.citeab.com/ycharos-antibody-characterization-data-citeab-search-engine/
46. 1. Lund-Johansen F

    (2023) A strong case for third-party testing

    eLife 12:e93329.

    https://doi.org/10.7554/eLife.93329

    • PubMed
    • Google Scholar
47. 1. Malički M

    (2024) Structure peer review to make it more robust

    Nature 628:483.

    https://doi.org/10.1038/d41586-024-01101-9

    • PubMed
    • Google Scholar
48. 1. Marcon E
    2. Jain H
    3. Bhattacharya A
    4. Guo H
    5. Phanse S
    6. Pu S
    7. Byram G
    8. Collins BC
    9. Dowdell E
    10. Fenner M
    11. Guo X
    12. Hutchinson A
    13. Kennedy JJ
    14. Krastins B
    15. Larsen B
    16. Lin ZY
    17. Lopez MF
    18. Loppnau P
    19. Miersch S
    20. Nguyen T
    21. Olsen JB
    22. Paduch M
    23. Ravichandran M
    24. Seitova A
    25. Vadali G
    26. Vogelsang MS
    27. Whiteaker JR
    28. Zhong G
    29. Zhong N
    30. Zhao L
    31. Aebersold R
    32. Arrowsmith CH
    33. Emili A
    34. Frappier L
    35. Gingras AC
    36. Gstaiger M
    37. Paulovich AG
    38. Koide S
    39. Kossiakoff AA
    40. Sidhu SS
    41. Wodak SJ
    42. Gräslund S
    43. Greenblatt JF
    44. Edwards AM

    (2015) Assessment of a method to characterize antibody selectivity and specificity for use in immunoprecipitation

    Nature Methods 12:725–731.

    https://doi.org/10.1038/nmeth.3472

    • PubMed
    • Google Scholar
49. 1. Marx V

    (2014) Cell-line authentication demystified

    Nature Methods 11:483–488.

    https://doi.org/10.1038/nmeth.2932

    • Google Scholar
50. 1. Menke J
    2. Roelandse M
    3. Ozyurt B
    4. Martone M
    5. Bandrowski A

    (2020) The Rigor and Transparency Index quality metric for assessing biological and medical science methods

    iScience 23:101698.

    https://doi.org/10.1016/j.isci.2020.101698

    • PubMed
    • Google Scholar
51. 1. Methods N

    (2015) Protein binder woes

    Nature Methods 12:373.

    https://doi.org/10.1038/nmeth.3397

    • Google Scholar
52. 1. Mitchell KG
    2. Gong B
    3. Hunter SS
    4. Burkart-Waco D
    5. Gavira-O’Neill CE
    6. Templeton KM
    7. Goethel ME
    8. Bzymek M
    9. MacNiven LM
    10. Murray KD
    11. Settles ML
    12. Froenicke L
    13. Trimmer JS

    (2023) High-volume hybridoma sequencing on the NeuroMabSeq platform enables efficient generation of recombinant monoclonal antibodies and scFvs for neuroscience research

    Scientific Reports 13:16200.

    https://doi.org/10.1038/s41598-023-43233-4

    • PubMed
    • Google Scholar
53. 1. Nilsson P
    2. Paavilainen L
    3. Larsson K
    4. Odling J
    5. Sundberg M
    6. Andersson AC
    7. Kampf C
    8. Persson A
    9. Al-Khalili Szigyarto C
    10. Ottosson J
    11. Björling E
    12. Hober S
    13. Wernérus H
    14. Wester K
    15. Pontén F
    16. Uhlen M

    (2005) Towards a human proteome atlas: High-throughput generation of mono-specific antibodies for tissue profiling

    Proteomics 5:4327–4337.

    https://doi.org/10.1002/pmic.200500072

    • PubMed
    • Google Scholar
54. 1. Omenn GS

    (2021) Reflections on the HUPO Human Proteome Project, the flagship project of the Human Proteome Organization, at 10 years

    Molecular & Cellular Proteomics 20:100062.

    https://doi.org/10.1016/j.mcpro.2021.100062

    • PubMed
    • Google Scholar
55. 1. Pillai-Kastoori L
    2. Heaton S
    3. Shiflett SD
    4. Roberts AC
    5. Solache A
    6. Schutz-Geschwender AR

    (2020) Antibody validation for Western blot: By the user, for the user

    Journal of Biological Chemistry 295:926–939.

    https://doi.org/10.1074/jbc.RA119.010472

    • PubMed
    • Google Scholar
56. 1. Polakiewicz RD

    (2015) Antibodies: The solution is validation

    Nature 518:483.

    https://doi.org/10.1038/518483b

    • PubMed
    • Google Scholar
57. 1. Roncador G
    2. Engel P
    3. Maestre L
    4. Anderson AP
    5. Cordell JL
    6. Cragg MS
    7. Šerbec VČ
    8. Jones M
    9. Lisnic VJ
    10. Kremer L
    11. Li D
    12. Koch-Nolte F
    13. Pascual N
    14. Rodríguez-Barbosa JI
    15. Torensma R
    16. Turley H
    17. Pulford K
    18. Banham AH

    (2016) The European antibody network’s practical guide to finding and validating suitable antibodies for research

    mAbs 8:27–36.

    https://doi.org/10.1080/19420862.2015.1100787

    • PubMed
    • Google Scholar
58. 1. Roy AL
    2. Wilder EL
    3. Anderson JM

    (2021) Validation of antibodies: Lessons learned from the Common Fund Protein Capture Reagents Program

    Science Advances 7:eabl7148.

    https://doi.org/10.1126/sciadv.abl7148

    • PubMed
    • Google Scholar
59. 1. Saper CB
    2. Sawchenko PE

    (2003) Magic peptides, magic antibodies: guidelines for appropriate controls for immunohistochemistry

    Journal of Comparative Neurology 465:161–163.

    https://doi.org/10.1002/cne.10858

    • PubMed
    • Google Scholar
60. 1. Saper CB

    (2005) An open letter to our readers on the use of antibodies

    Journal of Comparative Neurology 493:477–478.

    https://doi.org/10.1002/cne.20839

    • PubMed
    • Google Scholar
61. 1. Saper CB

    (2009) A guide to the perplexed on the specificity of antibodies

    Journal of Histochemistry and Cytochemistry 57:1–5.

    https://doi.org/10.1369/jhc.2008.952770

    • PubMed
    • Google Scholar
62. 1. Singh Chawla D

    (2015) Researchers argue for standard format to cite lab resources

    Nature.

    https://doi.org/10.1038/nature.2015.17652

    • Google Scholar
63. 1. Sivertsson Å
    2. Lindström E
    3. Oksvold P
    4. Katona B
    5. Hikmet F
    6. Vuu J
    7. Gustavsson J
    8. Sjöstedt E
    9. von Feilitzen K
    10. Kampf C
    11. Schwenk JM
    12. Uhlén M
    13. Lindskog C

    (2020) Enhanced validation of antibodies enables the discovery of missing proteins

    Journal of Proteome Research 19:4766–4781.

    https://doi.org/10.1021/acs.jproteome.0c00486

    • PubMed
    • Google Scholar
64. 1. Souren NY
    2. Fusenig NE
    3. Heck S
    4. Dirks WG
    5. Capes-Davis A
    6. Bianchini F
    7. Plass C

    (2022) Cell line authentication: A necessity for reproducible biomedical research

    EMBO Journal 41:e111307.

    https://doi.org/10.15252/embj.2022111307

    • PubMed
    • Google Scholar
65. 1. Stoevesandt O
    2. Taussig MJ

    (2012a) Affinity proteomics: the role of specific binding reagents in human proteome analysis

    Expert Review of Proteomics 9:401–414.

    https://doi.org/10.1586/epr.12.34

    • PubMed
    • Google Scholar
66. 1. Stoevesandt O
    2. Taussig MJ

    (2012b) European and international collaboration in affinity proteomics

    New Biotechnology 29:511–514.

    https://doi.org/10.1016/j.nbt.2012.05.003

    • PubMed
    • Google Scholar
67. 1. Taussig MJ
    2. Stoevesandt O
    3. Borrebaeck CAK
    4. Bradbury AR
    5. Cahill D
    6. Cambillau C
    7. de Daruvar A
    8. Dübel S
    9. Eichler J
    10. Frank R
    11. Gibson TJ
    12. Gloriam D
    13. Gold L
    14. Herberg FW
    15. Hermjakob H
    16. Hoheisel JD
    17. Joos TO
    18. Kallioniemi O
    19. Koegl M
    20. Konthur Z
    21. Korn B
    22. Kremmer E
    23. Krobitsch S
    24. Landegren U
    25. van der Maarel S
    26. McCafferty J
    27. Muyldermans S
    28. Nygren PA
    29. Palcy S
    30. Plückthun A
    31. Polic B
    32. Przybylski M
    33. Saviranta P
    34. Sawyer A
    35. Sherman DJ
    36. Skerra A
    37. Templin M
    38. Ueffing M
    39. Uhlén M

    (2007) ProteomeBinders: Planning a European resource of affinity reagents for analysis of the human proteome

    Nature Methods 4:13–17.

    https://doi.org/10.1038/nmeth0107-13

    • PubMed
    • Google Scholar
68. 1. Taussig MJ
    2. Fonseca C
    3. Trimmer JS

    (2018) Antibody validation: A view from the mountains

    New Biotechnology 45:1–8.

    https://doi.org/10.1016/j.nbt.2018.08.002

    • PubMed
    • Google Scholar
69. 1. Uhlén M
    2. Björling E
    3. Agaton C
    4. Szigyarto CAK
    5. Amini B
    6. Andersen E
    7. Andersson AC
    8. Angelidou P
    9. Asplund A
    10. Asplund C
    11. Berglund L
    12. Bergström K
    13. Brumer H
    14. Cerjan D
    15. Ekström M
    16. Elobeid A
    17. Eriksson C
    18. Fagerberg L
    19. Falk R
    20. Fall J
    21. Forsberg M
    22. Björklund MG
    23. Gumbel K
    24. Halimi A
    25. Hallin I
    26. Hamsten C
    27. Hansson M
    28. Hedhammar M
    29. Hercules G
    30. Kampf C
    31. Larsson K
    32. Lindskog M
    33. Lodewyckx W
    34. Lund J
    35. Lundeberg J
    36. Magnusson K
    37. Malm E
    38. Nilsson P
    39. Odling J
    40. Oksvold P
    41. Olsson I
    42. Oster E
    43. Ottosson J
    44. Paavilainen L
    45. Persson A
    46. Rimini R
    47. Rockberg J
    48. Runeson M
    49. Sivertsson A
    50. Sköllermo A
    51. Steen J
    52. Stenvall M
    53. Sterky F
    54. Strömberg S
    55. Sundberg M
    56. Tegel H
    57. Tourle S
    58. Wahlund E
    59. Waldén A
    60. Wan J
    61. Wernérus H
    62. Westberg J
    63. Wester K
    64. Wrethagen U
    65. Xu LL
    66. Hober S
    67. Pontén F

    (2005) A human protein atlas for normal and cancer tissues based on antibody proteomics

    Molecular & Cellular Proteomics 4:1920–1932.

    https://doi.org/10.1074/mcp.M500279-MCP200

    • PubMed
    • Google Scholar
70. 1. Uhlen M
    2. Bandrowski A
    3. Carr S
    4. Edwards A
    5. Ellenberg J
    6. Lundberg E
    7. Rimm DL
    8. Rodriguez H
    9. Hiltke T
    10. Snyder M
    11. Yamamoto T

    (2016) A proposal for validation of antibodies

    Nature Methods 13:823–827.

    https://doi.org/10.1038/nmeth.3995

    • PubMed
    • Google Scholar
71. 1. Vasilevsky NA
    2. Brush MH
    3. Paddock H
    4. Ponting L
    5. Tripathy SJ
    6. Larocca GM
    7. Haendel MA

    (2013) On the reproducibility of science: Unique identification of research resources in the biomedical literature

    PeerJ 1:e148.

    https://doi.org/10.7717/peerj.148

    • PubMed
    • Google Scholar
72. 1. Venkataraman A
    2. Yang K
    3. Irizarry J
    4. Mackiewicz M
    5. Mita P
    6. Kuang Z
    7. Xue L
    8. Ghosh D
    9. Liu S
    10. Ramos P
    11. Hu S
    12. Bayron Kain D
    13. Keegan S
    14. Saul R
    15. Colantonio S
    16. Zhang H
    17. Behn FP
    18. Song G
    19. Albino E
    20. Asencio L
    21. Ramos L
    22. Lugo L
    23. Morell G
    24. Rivera J
    25. Ruiz K
    26. Almodovar R
    27. Nazario L
    28. Murphy K
    29. Vargas I
    30. Rivera-Pacheco ZA
    31. Rosa C
    32. Vargas M
    33. McDade J
    34. Clark BS
    35. Yoo S
    36. Khambadkone SG
    37. de Melo J
    38. Stevanovic M
    39. Jiang L
    40. Li Y
    41. Yap WY
    42. Jones B
    43. Tandon A
    44. Campbell E
    45. Montelione GT
    46. Anderson S
    47. Myers RM
    48. Boeke JD
    49. Fenyö D
    50. Whiteley G
    51. Bader JS
    52. Pino I
    53. Eichinger DJ
    54. Zhu H
    55. Blackshaw S

    (2018) A toolbox of immunoprecipitation-grade monoclonal antibodies to human transcription factors

    Nature Methods 15:330–338.

    https://doi.org/10.1038/nmeth.4632

    • PubMed
    • Google Scholar
73. 1. Virk HS
    2. Rekas MZ
    3. Biddle MS
    4. Wright AKA
    5. Sousa J
    6. Weston CA
    7. Chachi L
    8. Roach KM
    9. Bradding P

    (2019) Validation of antibodies for the specific detection of human TRPA1

    Scientific Reports 9:.

    https://doi.org/10.1038/s41598-019-55133-7

    • PubMed
    • Google Scholar
74. 1. Voskuil JLA
    2. Bandrowski A
    3. Begley CG
    4. Bradbury ARM
    5. Chalmers AD
    6. Gomes AV
    7. Hardcastle T
    8. Lund-Johansen F
    9. Plückthun A
    10. Roncador G
    11. Solache A
    12. Taussig MJ
    13. Trimmer JS
    14. Williams C
    15. Goodman SL

    (2020) The Antibody Society’s antibody validation webinar series

    mAbs 12:1794421.

    https://doi.org/10.1080/19420862.2020.1794421

    • PubMed
    • Google Scholar
75. Website

    1. Waldron A

    (2022) Antibodies 101: Validation

    Accessed July 18, 2024.

    https://blog.addgene.org/antibodies-101-validation
76. 1. Williams M

    (2018) Reagent validation to facilitate experimental reproducibility

    Current Protocols in Pharmacology 81:e40.

    https://doi.org/10.1002/cpph.40

    • PubMed
    • Google Scholar

Decision letter after peer review:

Thank you for submitting your article ‘Antibody Characterization: A Historical Overview of Approaches to Enhancing Reproducibility in Biomedical Research and Ways Forward’ to eLife for consideration as a Feature Article. Your article has been reviewed by two peer reviewers, and the evaluation has been overseen by the Chief Magazine Editor of eLife (Dr Peter Rodgers). The following individuals involved in review of your submission have agreed to reveal their identity: Fridtjof Lund-Johansen and Dario Alessi.

The reviewers and editor have drafted this decision letter to help you prepare a revised submission.Reviewer #1:

Summary

The authors have written an historical background for the problems with the quality of research antibodies and describe initiatives taken to improve the situation. There has indeed been a very positive development during the past years, and a discussion of how this has happened is important. The focus is on the authors' own initiatives. This is natural, and I think many readers are unaware of wonderful resources such as the DSHB, Neuromab and NABOR and YCharOS.

Personally, I would prefer a more in-depth analysis of the current problems and more specific suggestions for how to solve them.

[1.1] Are there really 6 million antibodies on the market. How many have ever been sold or used in publications? As for the estimated financial loss of up to $1.8bn in the US per year: how confident are the authors of this estimate? People in the business have told me that most of their revenue comes from sales of a few top-selling products: it would be good if some of the authors who are based in industry could comment on this.

[1.2] The authors suggest that it is crucial to test all 6 million antibodies in multiple applications. I doubt that anyone thinks this is realistic or necessary. I would encourage them to come up with ideas for a more realistic strategy. There has to be a pre-selection step. What will it be and who will select? The next questions are who will test and how should the results be made available? My personal opinion is that money should be spent on making KO cells for as many genes as possible starting with the most widely studied proteins and make the cells available to all.

[1.3] Until now, the industry standard for primary screening has largely been western blotting. The implication is that we have discarded antibodies that bind to conformational epitopes. With KO cells, we can look beyond WB. How will that impact antibody development? The authors explain that there is a clear difference between the work involved in generating optimal antibody reagents and in characterizing them. I would argue that the two are closely linked and that you get what you screen for. I therefore find it interesting that Neuromab uses ELISA for the primary screen. I am sure that a lot of readers would be interested in learning more about their pipeline.

[1.4] The topic of webportals is important and deserves attention. A successful portal can have immense market power. The authors mention the Antibodypedia (from the Human Protein Atlas), the Antibody Registry, the Euromabnet and YCharOS. The initiatives are great, but to me it looks as if CiteAb has already established a position as the analogue of Clarivate Analytics for antibodies. The authors may disagree, and the landscape can change quickly in the era of ChatGPT. Still, I don't think it is wise to ignore CiteAb's role in the current market.Reviewer #2:

Summary

This is an important review that comprehensively discusses approaches to enhancing the reproducibility of antibody characterization. The authors are the key researchers working in this area from researchers to funding bodies and foundations to antibody vendors.

The authors highlight the huge amounts of resources that are lost each year and the resulting poor science that is undertaken leading to unreproducible results from using antibodies that are poorly characterised and have not been sufficiently validated. The study provides a very good historical context, discussing many previous past efforts and more recent efforts to improve the rigour with which antibodies are characterised. These include the Human Protein Atlas, the Neuromab Facility, The Protein Capture Reagent Programme (PCRP), Affinomics, an EU-funded programme. The authors also discuss in depth the ‘Research Resource Identifier’ (RRID) Programme that provides a unique identifier to each antibody.

The authors also highlight the Developmental Studies Hybridoma Bank (DSHB) that has played such an important role in distributing hybridoma clones and more recently recombinant antibody cDNA vectors. The authors also discussed the Clinical Proteome Tumour Analysis Consortium (CPTAC) that provides antibodies for cancer research. The authors also discuss a very valuable effort, started by the Structural Genome Consortium, called the Antibody Characterisation Through Open Science (YCharOS). This consortium have rigorously tested over 1000 antibodies and characterised detailed antibody reports on many of these, and have found that a significant proportion of all antibodies that have been used in many previous publications shockingly do not detect endogenous proteins.

The authors also discuss a community called Only Good Antibodies (OGA) that are also attempting to validate antibodies in a consistent manner and provide this information to the research community. The authors also highlight the important roles that researchers, vendors, universities, journals, funding society and disease foundations can have to mandate that researchers only use well characterised antibodies, or if the antibody hasn't been well characterised previously, the authors need to undertake essential characterisation, demonstrating that the antibodies they are using are indeed recognising the designated target, preferably using a knockout or knockdown approach. The authors also highlight important work undertaken by the Michael J. Fox Foundation for Parkinson's research where they've developed a unique “research tools programme” which focuses on characterisation and validation of antibodies used for research. Thus far, 200 research tools have been produced to date that are all highly characterised, and groups using these reagents can be reassured that the antibodies are very well characterised.

The authors also discuss the importance of trying to move away from polyclonal antibodies to recombinant antibodies in which the sequences are known and antibodies can be expressed more consistently in human cell lines. Ideally having sequences available for antibodies certainly helps researchers to ensure that the antibody is the correct reagent and enables them to express these reagents as scale in their laboratories, rather than having to purchase these critical reagemts at high cost from a vendor.

The authors provide a lot of important suggestions and recommendations including improving awareness and training in laboratories of the importance of rigorous antibody characterization has in ensuring the reproducibility of life sciences research.

This article is a must read for Life Scientists at any stages of their careers who make use of antibodies for their research. I recommend acceptance of this study.

[2.1] The authors highlight the major problem with companies that distribute antibodies to multiple other companies resulting in the same antibodies having different RRID numbers, that causes major confusion. It would be important to put in place a system whereby if a company purchases an antibody from another vendor that they tried to ensure that the original RRID number is used for this purpose-this may be difficult to achieve. The authors should comment on this.

Linking antibodies with their sequences would also be good but also unlikely to work as most antibody companies do not disclose the sequences of their antibodies for commercial reasons. It would be good if the authors could also comment on this.

[2.2] YCharOS found that on average 12 publications per protein target contain data from an antibody that failed to recognise the endogenous targets. YCharOS is an extremely important effort and I hope that it will be able to continue expanding their efforts to properly validate many of the widely used antibodies. This effort represents a drop in the ocean and the feasibility of dramatically scalling up the efforts of YCharOS to test a greater proportion of all widely used antibodies seems very challenging.

It would be good if the authors could say more about the challenges involved in scaling up such efforts.

[2.3] It would be helpful to have a table or figure outlining the key steps/criteria that the authors believe are needed to thoroughly characterise antibodies for immunoblotting, immunofluorescence, immunoprecipitation, immunohistochemistry, and flow cytometry. This might help students and postdocs plan their experiments to ensure that the data required to validate the antibodies that they're using is available. Ideally, vendors would also sign up to this level of characterisation data being provided in the antibody datasheets. Ideally third-party companies or consortia could be available to undertake this important QC work in the future and feed the back to the community.

[2.4] If the sequences of the recombinant antibody are known, using Alphafold 3 it should be possible to predict the structure of the antibody antigen complex. This might enable researchers to generate subtle point mutations in the recombinant antibody that ablates their affinity for the intended target. Such mutated antibodies could serve as useful control reagents for certain applications such as IF, IHC, flow cytometry, IP, to ensure that the signals that are observed are indeed mediated by binding to the antigen and not to a non-specific site.

It would be good if the authors could comment on this (including the possibility that vendors might start providing these control reagents in the future).

[2.5] An important recommendation that is not made in the review (perhaps because vendors are co-authors) is that vendors should sign up to withdraw all antibodies found not to work or detect the endogenous target by trusted third party testers or consortia. Again it would be good to discuss this issue in the review.

Reviewer #1:

Summary

The authors have written an historical background for the problems with the quality of research antibodies and describe initiatives taken to improve the situation. There has indeed been a very positive development during the past years, and a discussion of how this has happened is important. The focus is on the authors' own initiatives. This is natural, and I think many readers are unaware of wonderful resources such as the DSHB, Neuromab and NABOR and YCharOS.

Personally, I would prefer a more in-depth analysis of the current problems and more specific suggestions for how to solve them.

We actually agree that a more in-depth analysis of the problems and solutions would be valuable. However, the number and complexities of the issues forces us to limit the depth attainable in a review that itself is limited in length. Several of the topics covered have been the subject of articles that focus on specific issues and are covered in more depth, and we cite these whenever possible, to allow readers to find greater depth of discussion of specific issues. One goal of the article was to emphasize the need for community efforts to work together to move towards solutions for the problems so it did not seem prudent to get too detailed in telling each group what they should do ahead of time.

[1.1] Are there really 6 million antibodies on the market. How many have ever been sold or used in publications? As for the estimated financial loss of up to $1.8bn in the US per year: how confident are the authors of this estimate? People in the business have told me that most of their revenue comes from sales of a few top-selling products: it would be good if some of the authors who are based in industry could comment on this.

Yes, according to industry sources there are about 6 million antibodies on the market. CiteAb is an online source that tracks antibodies and other reagents and they report that their database currently covers 7.4 million antibodies, though this number is clouded by the (unknown) numbers of instances in which different companies are selling the same antibody, typically with different identifiers. The Antibody Registry lists around 3 million antibodies. The Antibody Registry listings largely miss companies that are newly created and those that are in the business of aggregating and reselling antibodies from other companies; e.g., those that are targeting sales in a country specific manner, thus this might be a lower boundary. These numbers are also growing steadily in both databases.

As to the financial losses in the US, estimates have varied and we acknowledge that they are just that, estimates. We have now included more citations and include a range of estimates from those publications, rather than just the upper end that we included previously. Note that the estimates are larger in the more recent publications. Even the lowest estimate should be concerning to all readers.

It is true that most revenue from sales of antibodies comes from a relatively few top selling products. These can be found on CiteAb, along with other metrics. Reviewer comments have helped us identify CiteAb as a valuable resource that we now include in several places in the article (see reviewer comment and response below).

[1.2] The authors suggest that it is crucial to test all 6 million antibodies in multiple applications. I doubt that anyone thinks this is realistic or necessary. I would encourage them to come up with ideas for a more realistic strategy. There has to be a pre-selection step. What will it be and who will select? The next questions are who will test and how should the results be made available? My personal opinion is that money should be spent on making KO cells for as many genes as possible starting with the most widely studied proteins and make the cells available to all.

We are sorry if we left the impression that we are arguing that all commercial antibodies should be tested. Rather, we think an achievable, though still distant, goal is to identify and characterize antibodies targeting each protein in proteomes of at least the major model organisms (human, mouse, rat, worms, flies, yeast), focusing first on humans. So rather than pick an antibody and then characterize it, the approach needs to be to identify a protein and then test all available, unique antibodies (a difficult task as many resellers offer the same product under different catalog numbers, clone numbers and may even crop and change contrast on characterization data) and make the data publicly available (YCharOs approach), and in the process eliminating the bad ones and updating available information to allow researchers to make the most informed choices in purchasing their reagents. Those proteins which lack good antibodies can then be targets for antibody discovery.

In response to this comment, we have also added some additional, specific and realistic strategies that can be undertaken or expanded upon in the near future. Among these, we have expanded upon the idea of small groups of experts in a field working together, under the ‘Research scientists/end users’ section.

There already is a pre-selection step in the antibody characterization work currently underway; selection is dictated by available funding. Federal funding agencies are currently resistant to supporting such efforts broadly (though the article points out previous supported projects and highlights why many have fallen short of expectations) and this appears unlikely to change in the near future. This is also why it is important to highlight support from other funding sources. To further emphasize this point we have added some text and a reference (Davies, et al. (2013) Biochem J) that gives a great example of what can be achieved with such support. Given the lack of interest by NIH leadership in funding this work there seems little reason to get into the different ways one might prioritize pre-selection steps, although it is already being done by foundations that support such efforts.

We also strongly agree on the value of KO cells/tissues/model organisms in this arena and of making them readily available. There is currently no resource we could find for depositing and sharing KO cell lines (though some model organisms (mice, yeast, worms, flies) do have the capability to deposit and share mutants). This is an issue worthy of additional discussion and we have included text pointing out the lack of this potentially valuable resource in the article.

[1.3] Until now, the industry standard for primary screening has largely been western blotting. The implication is that we have discarded antibodies that bind to conformational epitopes. With KO cells, we can look beyond WB. How will that impact antibody development? The authors explain that there is a clear difference between the work involved in generating optimal antibody reagents and in characterizing them. I would argue that the two are closely linked and that you get what you screen for. I therefore find it interesting that Neuromab uses ELISA for the primary screen. I am sure that a lot of readers would be interested in learning more about their pipeline.

We completely agree with the issues raised in this comment; e.g., that how screening is done can influence the outcome and bias antibodies that work in certain assays over others. It is not clear how the availability of KO cell lines will alter antibody development but we emphasize throughout the article their value in characterization efforts. The issue of relying on ELISA as a first screen is a complicated one as there is not always good correlation between ELISA positive antibodies and ones that work well in other assays (as pointed out in the cited article from Gong, et al), though the fact that it is rapid, scalable, and relatively cheap means it will likely remain the industry standard at least for now. In response to the request for more details on the NeuroMab screening we have increased details in this section. We have also discussed briefly the issue of ELISA vs other assays as well as antibody production involving animals vs displays. Note that we do not go into details or comparisons of ways to generate and screen for antibodies but try to remain focused on the complicated enough issues surrounding their characterization.

[1.4] The topic of webportals is important and deserves attention. A successful portal can have immense market power. The authors mention the Antibodypedia (from the Human Protein Atlas), the Antibody Registry, the Euromabnet and YCharOS. The initiatives are great, but to me it looks as if CiteAb has already established a position as the analogue of Clarivate Analytics for antibodies. The authors may disagree, and the landscape can change quickly in the era of ChatGPT. Still, I don't think it is wise to ignore CiteAb's role in the current market.

Point well taken. We have reached out to CiteAb to get information on their activities and history and have now included a paragraph describing that work. It was also sent to the leadership at CiteAb to ensure accuracy and as a result of their efforts to write/edit and ensure the accuracy of our write up, a representative from CiteAb has been included as a co-author. We note that while a potentially useful resource they use data mining from references predominantly so risk exacerbating the problem of using numbers of references as evidence of antibody characterization. As a result we have also included citations to a couple of their online editorials, including ones that do a good job of elaborating on this issue as well as one walking the reader through steps proposed to find a good antibody.

Reviewer #2:

Summary

This is an important review that comprehensively discusses approaches to enhancing the reproducibility of antibody characterization. The authors are the key researchers working in this area from researchers to funding bodies and foundations to antibody vendors.

The authors highlight the huge amounts of resources that are lost each year and the resulting poor science that is undertaken leading to unreproducible results from using antibodies that are poorly characterised and have not been sufficiently validated. The study provides a very good historical context, discussing many previous past efforts and more recent efforts to improve the rigour with which antibodies are characterised. These include the Human Protein Atlas, the Neuromab Facility, The Protein Capture Reagent Programme (PCRP), Affinomics, an EU-funded programme. The authors also discuss in depth the ‘Research Resource Identifier’ (RRID) Programme that provides a unique identifier to each antibody.

The authors also highlight the Developmental Studies Hybridoma Bank (DSHB) that has played such an important role in distributing hybridoma clones and more recently recombinant antibody cDNA vectors.

We want to clarify that DSHB distributes recombinant and monoclonal antibodies but not vectors that may be used to express them. We have gone over the text to ensure this point is clear.

The authors also discussed the Clinical Proteome Tumour Analysis Consortium (CPTAC) that provides antibodies for cancer research. The authors also discuss a very valuable effort, started by the Structural Genome Consortium, called the Antibody Characterisation Through Open Science (YCharOS). This consortium have rigorously tested over 1000 antibodies and characterised detailed antibody reports on many of these, and have found that a significant proportion of all antibodies that have been used in many previous publications shockingly do not detect endogenous proteins.

The authors also discuss a community called Only Good Antibodies (OGA) that are also attempting to validate antibodies in a consistent manner and provide this information to the research community. The authors also highlight the important roles that researchers, vendors, universities, journals, funding society and disease foundations can have to mandate that researchers only use well characterised antibodies, or if the antibody hasn't been well characterised previously, the authors need to undertake essential characterisation, demonstrating that the antibodies they are using are indeed recognising the designated target, preferably using a knockout or knockdown approach. The authors also highlight important work undertaken by the Michael J. Fox Foundation for Parkinson's research where they've developed a unique "research tools programme" which focuses on characterisation and validation of antibodies used for research. Thus far, 200 research tools have been produced to date that are all highly characterised, and groups using these reagents can be reassured that the antibodies are very well characterised.

The authors also discuss the importance of trying to move away from polyclonal antibodies to recombinant antibodies in which the sequences are known and antibodies can be expressed more consistently in human cell lines. Ideally having sequences available for antibodies certainly helps researchers to ensure that the antibody is the correct reagent and enables them to express these reagents as scale in their laboratories, rather than having to purchase these critical reagemts at high cost from a vendor.

The authors provide a lot of important suggestions and recommendations including improving awareness and training in laboratories of the importance of rigorous antibody characterization has in ensuring the reproducibility of life sciences research.

This article is a must read for Life Scientists at any stages of their careers who make use of antibodies for their research. I recommend acceptance of this study.

[2.1] The authors highlight the major problem with companies that distribute antibodies to multiple other companies resulting in the same antibodies having different RRID numbers, that causes major confusion. It would be important to put in place a system whereby if a company purchases an antibody from another vendor that they tried to ensure that the original RRID number is used for this purpose-this may be difficult to achieve. The authors should comment on this.

Linking antibodies with their sequences would also be good but also unlikely to work as most antibody companies do not disclose the sequences of their antibodies for commercial reasons. It would be good if the authors could also comment on this.

Both great suggestions/comments. There are ongoing discussions among stakeholders to develop a better system in which each antibody is given one and only one RRID number. Those involved admit this will be difficult to enforce and track but all agree it is a worthwhile effort. We have included this in the article as the worthy goal that it is.

Under the RRID section we have added:

“It is also important to note that one antibody can be sold by multiple different vendors with each using a different RRID number, and different lots of the same manufacturer’s antibody will have the same RRID, even if there may be significant lot to lot variation. To the extent possible, this practice should be replaced by better practices, including assigning one RRID to a reagent so that each antibody is given one and only one RRID number.”

Under Antibody vendors and repositories we have added this:

“Vendors should also take the lead in ensuring that each antibody is assigned one, and only one, RRID number to allow better tracking and linkage to characterization data. When distributing to other vendors they should have the opportunity to make this a requirement.”

As to the linkage of sequences to antibodies, you are correct that vendors will not do so. This is in response to experience in which other companies, typically overseas, have used the sequence information to generate the same product and sell it without compensation to the originator. While a worthy goal, we believe it is unrealistic today, though now include the following statement in the article, under the NeuroMab section:

“The mAbs and hybridomas that produce them are distributed through the DSHB and the rAb sequences and plasmids are available at Addgene. It is worth noting here both the value in making sequences of antibodies available to researchers but also the limitations on vendors, as commercial enterprises, who cannot readily do so without risking use of the information by competitors.”

[2.2] YCharOS found that on average 12 publications per protein target contain data from an antibody that failed to recognise the endogenous targets. YCharOS is an extremely important effort and I hope that it will be able to continue expanding their efforts to properly validate many of the widely used antibodies. This effort represents a drop in the ocean and the feasibility of dramatically scalling up the efforts of YCharOS to test a greater proportion of all widely used antibodies seems very challenging.

It would be good if the authors could say more about the challenges involved in scaling up such efforts.

As with so many things in research, the challenges in scale up get back to funding. This is a theme we try to bring out throughout the article and hope that federal funding agencies are paying attention. YCharOS is in the process of scaling up, with sites in Canada and the UK currently and a third planned in the US. We also believe that groups of experts in each field are ideal for shared efforts in this arena and may allow for funding in more focused projects, targeting perhaps all members of a protein family or the known and emerging players in a specific pathology or pathway. The practical limitation of scaling up efforts is that among the most common assays used by researchers (WB and IHC) are not readily scalable or automated without migrating to a platform that may not sufficiently resemble what is done in the average research laboratory.

[2.3] It would be helpful to have a table or figure outlining the key steps/criteria that the authors believe are needed to thoroughly characterise antibodies for immunoblotting, immunofluorescence, immunoprecipitation, immunohistochemistry, and flow cytometry. This might help students and postdocs plan their experiments to ensure that the data required to validate the antibodies that they're using is available. Ideally, vendors would also sign up to this level of characterisation data being provided in the antibody datasheets. Ideally third-party companies or consortia could be available to undertake this important QC work in the future and feed the back to the community.

We agree and found this to be an excellent suggestion that we are happy to comply with. We generated a new Figure (Figure 1) that summarizes these issues and hope it helps clarify in the minds of students and all end users the issues and steps that can be taken.

In addition, YCharOS has recently submitted a manuscript, currently under review at a leading Methods journal, with step-by-step protocols for WB/IP/IF and put it up on Protocols Exchange, a preprint server. Representatives from 10 leading antibody manufacturers are co-authors on this article, demonstrating the adoption of these protocols by the industry. Also, NeuroMab has made their IHC protocols openly available.

For reasons included in the article, it remains financially unsupportable for vendors to perform the characterization work comparable to YCharOS protocols as costs for analyses of commercial antibodies targeting one protein may be estimated at ~$30,000, which is far more than a company makes from selling the average antibody.

[2.4] If the sequences of the recombinant antibody are known, using Alphafold 3 it should be possible to predict the structure of the antibody antigen complex. This might enable researchers to generate subtle point mutations in the recombinant antibody that ablates their affinity for the intended target. Such mutated antibodies could serve as useful control reagents for certain applications such as IF, IHC, flow cytometry, IP, to ensure that the signals that are observed are indeed mediated by binding to the antigen and not to a non-specific site.

It would be good if the authors could comment on this (including the possibility that vendors might start providing these control reagents in the future).

This too is an excellent point and we have included the future potential for deep learning algorithms like AlphaFold to play in this field, in the section headed Technology Development. We also used this as another argument in support of vendors making sequence data readily available, though for reasons already given this is unlikely to become standard practice in the industry.

The newly added text:

“Finally, a future role for technology in this field that will likely play increasingly important roles in the development and optimization of antibodies will come from deep learning models, such as AlphaFold. These models should enable predictions of the antibody-antigen complex, aid in the identification of the epitope targeted, and help determine if folding, post-translational modifications or other issues may influence the output from use of the antibody. These future prospects are another reason to encourage open access to antibody sequences of rAbs.”

The use of AlphaFold to generate point mutations for use as negative controls was not deemed as useful for a number of reasons including: (1) modified antibodies could develop specificity for other unwanted proteins and (2) the use of WT versus modified antibodies won’t confirm whether the signal from the WT antibody was coming from the intended target protein or unwanted proteins.

[2.5] An important recommendation that is not made in the review (perhaps because vendors are co-authors) is that vendors should sign up to withdraw all antibodies found not to work or detect the endogenous target by trusted third party testers or consortia. Again it would be good to discuss this issue in the review.

Responsible vendors have already stepped up and are taking appropriate steps in response to data becoming available. This is best demonstrated by the fact that the YCharOS manufacturer partners have removed 20% of antibodies tested by YCharOS that failed in testing, and 40% of antibodies have had their recommendations changed (Ayoubi et al., eLife, 2023). We also point out again that just because an antibody does not perform well in one assay does not mean it won’t be useful in a different use or in the same assay but under different conditions. So it is not reasonable to insist on withdrawal of any antibody, except those found to not bind the stated target.

In summary, we again thank the reviewers for their input and believe we have responded appropriately to each of their comments/suggestions and as a result believe the manuscript should be ready for publication.

Author details

1. Richard A Kahn

   Richard A Kahn is in the Department of Biochemistry, Emory University School of Medicine, Atlanta, United States

   Contribution

   Conceptualization, Writing – original draft, Project administration, Writing – review and editing

   For correspondence

   rkahn@emory.edu

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-0259-0601

2. Harvinder Virk

   Harvinder Virk is in the Department of Respiratory Sciences, University of Leicester, Leicester, United Kingdom

   Contribution

   Conceptualization, Writing – original draft, Writing – review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-9739-9593

3. Carl Laflamme

   Carl Laflamme is in the Department of Neurology and Neurosurgery, Structural Genomics Consortium, The Montreal Neurological Institute, McGill University, Montreal, Canada

   Contribution

   Conceptualization, Writing – original draft, Writing – review and editing, Drafted the section on YCharOS

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-5906-025X

4. Douglas W Houston

   Douglas W Houston is at the Development Studies Hybridoma Databank, University of Iowa, Iowa City, United States

   Contribution

   Writing – original draft, Writing – review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-7438-4655

5. Nicole K Polinski

   Nicole K Polinski is at The Michael J Fox Foundation for Parkinson’s Research, New York, United States

   Contribution

   Writing – original draft, Writing – review and editing, Drafted the section on the Michael J Fox Foundation

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-4613-9115

6. Rob Meijers

   Rob Meijers is at the Institute for Protein Innovation, Boston, United States

   Contribution

   Writing – original draft, Writing – review and editing

   Competing interests

   No competing interests declared

7. Allan I Levey

   Allan I Levey is in the Department of Neurology, Emory University School of Medicine, Atlanta, United States

   Contribution

   Writing – review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-3153-502X

8. Clifford B Saper

   Clifford B Saper is in the Department of Neurology and Program in Neuroscience, Harvard Medical School and Beth Israel Deaconess Medical Center, Boston, United States

   Contribution

   Writing – review and editing

   Competing interests

   No competing interests declared

9. Timothy M Errington

   Timothy M Errington is at the Center for Open Science, Charlottesville, United States

   Contribution

   Writing – review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-4959-5143

10. Rachel E Turn

    Rachel E Turn is in the Department of Microbiology and Immunology, Stanford University School of Medicine, Stanford, United States

    Contribution

    Writing – review and editing

    Competing interests

    No competing interests declared

    "This ORCID iD identifies the author of this article:" 0000-0001-5389-4560

11. Anita Bandrowski

    Anita Bandrowski is in the Department of Neuroscience, University of California at San Diego, La Jolla, United States

    Contribution

    Writing – review and editing

    Competing interests

    Co-founder and serving as CEO of SciCrunch Inc, a company that works with publishers to improve the scientific literature; the terms of this agreement have been approved by the COI office at the University of California at San Diego

    "This ORCID iD identifies the author of this article:" 0000-0002-5497-0243

12. James S Trimmer

    James S Trimmer is in the Department of Physiology and Membrane Biology, University of California Davis School of Medicine, Davis, United States

    Contribution

    Writing – review and editing, Drafted the section on NeuroMab/NABOR

    Competing interests

    No competing interests declared

    "This ORCID iD identifies the author of this article:" 0000-0002-6117-3912

13. Meghan Rego

    Meghan Rego is at Addgene, Watertown, United States

    Contribution

    Writing – review and editing, Drafted the section on Addgene

    Competing interests

    Employed by Addgene, a company that may be affected financially by the opinions reported in this article

14. Leonard P Freedman

    Leonard P Freedman is in the Frederick National Laboratory for Cancer Research, Frederick, United States

    Contribution

    Writing – review and editing

    Competing interests

    No competing interests declared

    "This ORCID iD identifies the author of this article:" 0009-0007-5726-150X

15. Fortunato Ferrara

    Fortunato Ferrara is at Specifica, a Q² company, Santa Fe, United States

    Contribution

    Visualization, Writing – review and editing

    Competing interests

    No competing interests declared

16. Andrew RM Bradbury

    Andrew RM Bradbury is at Specifica, a Q² company, Santa Fe, United States

    Contribution

    Writing – review and editing

    Competing interests

    No competing interests declared

17. Hannah Cable

    Hannah Cable is in the Department of Research and Development, Abcam, Cambridge, United Kingdom

    Contribution

    Writing – review and editing

    Competing interests

    Employed by Abcam, a for-profit company that sells antibodies and which may be affected financially by the opinions reported in this article

    "This ORCID iD identifies the author of this article:" 0009-0001-0971-5761

18. Skye Longworth

    Skye Longworth is at CiteAb, Bath, United Kingdom

    Contribution

    Writing – review and editing

    Competing interests

    No competing interests declared