Active machine learning-driven experimentation to determine compound effects on protein patterns

  1. Armaghan W Naik
  2. Joshua D Kangas
  3. Devin P Sullivan
  4. Robert F Murphy  Is a corresponding author
  1. Carnegie Mellon University, United States

Abstract

High throughput screening determines the effects of many conditions on a given biological target. Currently, to estimate the effects of those conditions on other targets requires either strong modeling assumptions (e.g. similarities among targets) or separate screens. Ideally, data-driven experimentation could be used to learn accurate models for many conditions and targets without doing all possible experiments. We have previously described an active machine learning algorithm that can iteratively choose small sets of experiments to learn models of multiple effects. We now show that, with no prior knowledge and with liquid handling robotics and automated microscopy under its control, this learner accurately learned the effects of 48 chemical compounds on the subcellular localization of 48 proteins while performing only 29% of all possible experiments. The results represent the first practical demonstration of the utility of active learning-driven biological experimentation in which the set of possible phenotypes is unknown in advance.

Article and author information

Author details

  1. Armaghan W Naik

    Computational Biology Department, Center for Bioimage Informatics, Carnegie Mellon University, Pittsburgh, United States
    Competing interests
    The authors declare that no competing interests exist.
  2. Joshua D Kangas

    Computational Biology Department, Center for Bioimage Informatics, Carnegie Mellon University, Pittsburgh, United States
    Competing interests
    The authors declare that no competing interests exist.
  3. Devin P Sullivan

    Computational Biology Department, Center for Bioimage Informatics, Carnegie Mellon University, Pittsburgh, United States
    Competing interests
    The authors declare that no competing interests exist.
  4. Robert F Murphy

    Computational Biology Department, Center for Bioimage Informatics, Carnegie Mellon University, Pittsburgh, United States
    For correspondence
    murphy@cmu.edu
    Competing interests
    The authors declare that no competing interests exist.

Copyright

© 2016, Naik et al.

This article is distributed under the terms of the Creative Commons Attribution License permitting unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 7,115
    views
  • 1,237
    downloads
  • 43
    citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Armaghan W Naik
  2. Joshua D Kangas
  3. Devin P Sullivan
  4. Robert F Murphy
(2016)
Active machine learning-driven experimentation to determine compound effects on protein patterns
eLife 5:e10047.
https://doi.org/10.7554/eLife.10047

Share this article

https://doi.org/10.7554/eLife.10047

Further reading

    1. Cell Biology
    2. Developmental Biology
    Pavan K Nayak, Arul Subramanian, Thomas F Schilling
    Research Article Updated

    Mechanical forces play a critical role in tendon development and function, influencing cell behavior through mechanotransduction signaling pathways and subsequent extracellular matrix (ECM) remodeling. Here, we investigate the molecular mechanisms by which tenocytes in developing zebrafish embryos respond to muscle contraction forces during the onset of swimming and cranial muscle activity. Using genome-wide bulk RNA sequencing of FAC-sorted tenocytes we identify novel tenocyte markers and genes involved in tendon mechanotransduction. Embryonic tendons show dramatic changes in expression of matrix remodeling associated 5b (mxra5b), matrilin 1 (matn1), and the transcription factor kruppel-like factor 2a (klf2a), as muscles start to contract. Using embryos paralyzed either by loss of muscle contractility or neuromuscular stimulation we confirm that muscle contractile forces influence the spatial and temporal expression patterns of all three genes. Quantification of these gene expression changes across tenocytes at multiple tendon entheses and myotendinous junctions reveals that their responses depend on force intensity, duration, and tissue stiffness. These force-dependent feedback mechanisms in tendons, particularly in the ECM, have important implications for improved treatments of tendon injuries and atrophy.

    1. Cancer Biology
    2. Cell Biology
    Ida Marie Boisen, Nadia Krarup Knudsen ... Martin Blomberg Jensen
    Research Article

    Testicular microcalcifications consist of hydroxyapatite and have been associated with an increased risk of testicular germ cell tumors (TGCTs) but are also found in benign cases such as loss-of-function variants in the phosphate transporter SLC34A2. Here, we show that fibroblast growth factor 23 (FGF23), a regulator of phosphate homeostasis, is expressed in testicular germ cell neoplasia in situ (GCNIS), embryonal carcinoma (EC), and human embryonic stem cells. FGF23 is not glycosylated in TGCTs and therefore cleaved into a C-terminal fragment which competitively antagonizes full-length FGF23. Here, Fgf23 knockout mice presented with marked calcifications in the epididymis, spermatogenic arrest, and focally germ cells expressing the osteoblast marker Osteocalcin (gene name: Bglap, protein name). Moreover, the frequent testicular microcalcifications in mice with no functional androgen receptor and lack of circulating gonadotropins are associated with lower Slc34a2 and higher Bglap/Slc34a1 (protein name: NPT2a) expression compared with wild-type mice. In accordance, human testicular specimens with microcalcifications also have lower SLC34A2 and a subpopulation of germ cells express phosphate transporter NPT2a, Osteocalcin, and RUNX2 highlighting aberrant local phosphate handling and expression of bone-specific proteins. Mineral disturbance in vitro using calcium or phosphate treatment induced deposition of calcium phosphate in a spermatogonial cell line and this effect was fully rescued by the mineralization inhibitor pyrophosphate. In conclusion, testicular microcalcifications arise secondary to local alterations in mineral homeostasis, which in combination with impaired Sertoli cell function and reduced levels of mineralization inhibitors due to high alkaline phosphatase activity in GCNIS and TGCTs facilitate osteogenic-like differentiation of testicular cells and deposition of hydroxyapatite.