Humans age in the sense that an individual’s probability of dying increases as a function of time lived. Interestingly, this trend is not true of all species but is true of mammals generally (Jones et al., 2014). Numerous hypotheses of mammal-specific aging exist, including the antagonistic pleiotropy hypothesis (Williams, 1957) and the mutation accumulation hypothesis (Medawar, 1952), both of which refer to changes within an individual throughout its lifetime that result in aging. The antagonistic pleiotropy hypothesis postulates that genes that are beneficial to early life become detrimental later in life. Such genes are retained because they increase early-life reproductive output and thus increase fitness. The mutation accumulation hypothesis predicts that a gradual accumulation of errors in DNA sequence as a result of repeated replication during a lifetime’s worth of cell divisions will lead to a gradual breakdown of functionality. Support for both hypotheses has been found in individual species. Recent work has identified disease-related SNPs in age-related genes that are beneficial in early life and detrimental in later life in humans, thus indicating selective pressures associated with gene evolution related to aging and supporting the antagonistic pleiotropy hypothesis (Rodríguez et al., 2017). A study of SNVs in the human brain found that number of mutations was positively correlated with age and that mutations were at loci associated with age-related disease, thus supporting the mutation accumulation hypothesis (Lodato et al., 2018). However, in human populations, variability in the aging phenotype is limited and many confounding biological changes are correlated with aging, making it difficult to pinpoint specific biological processes that are causal and thus amenable to manipulation.

On the other hand, lifespan varies dramatically (>100 fold) across mammals (Nowak, 1999), making comparative genomics a fruitful avenue for aging research. Numerous studies have investigated the genomic features of mammals with extreme lifespan such as bats (Foley et al., 2018; Seim et al., 2013), naked mole-rats (Kim et al., 2011), whales (Keane et al., 2015), and elephants (Sulak et al., 2016) to identify potential causative genetic changes. In Myotis, the longest-lived bat genus, species show lack of telomere shortening and corresponding expression changes in telomere maintenance and DNA repair genes (Foley et al., 2018). Comparative genomics studies have also suggested that changes in the insulin growth factor one pathway may enable increased lifespan and cancer resistance in Myotis brandtii (Brandt’s bat) (Seim et al., 2013). Similarly, naked mole-rats show differential regulation of genes associated with macromolecule degradation, mitochondrial function, and TERT, a gene associated with telomere maintenance, as well as changes to genes related to tumor suppression (Kim et al., 2011). In primates, genes associated with cardiovascular function, coagulation, and healing have been demonstrated to show evolutionary correlations with lifespan (Muntané et al., 2018). Sequencing of the bowhead whale genome revealed species-specific changes in DNA repair, cell cycle, and aging genes (Keane et al., 2015). In elephants, a striking increase in TP53 copy number has been linked to increased cancer resistance enabling longer lifespan (Sulak et al., 2016). Despite compelling results, these single- and limited-species studies have limitations. The species studied differ from their nearest sequenced relatives in multiple physiological traits as well as millions of nucleotides. Thus, while single-species studies have yielded some credible candidates for genes associated with increased lifespan, it is difficult to know to what extent these represent insights into the universal mechanisms of lifespan regulation rather than species-specific adaptation or coincidental neutral changes. In this study, we develop new methodology to evaluate the relationship between the evolutionary constraint of genes and pathways and quantitative lifespan traits in an unbiased, genome-wide, pan-mammalian analysis.

The wide range of lifespans across the mammalian phylogeny (Figure 1A) provides the ideal dataset to investigate lifespan from a comparative genomics perspective. Because independent changes in lifespan occurred repeatedly in the mammalian species tree, lifespan can be viewed as a convergent trait. Molecular features that correlate with convergent changes in lifespan therefore may also occur repeatedly across a variety of organisms. In our study we use protein evolutionary rates quantified as the number of amino acid substitutions on a phylogenetic branch to infer convergent rate shifts associated with lifespan traits across the mammalian phylogeny.

Figure 1 with 11 supplements see all

Download asset Open asset

Evolutionary rates are useful for linking phenotypes to genes because they reflect evolutionary constraint experienced by a genetic element (Zhang and Yang, 2015). In the absence of diversifying selection, genetic elements that support a specific trait are expected to be more constrained in species where the trait has a larger contribution to fitness. In agreement with this expectation, multiple studies have shown that a genetic element providing a function less important for a given species is under less constraint and hence exhibits a faster evolutionary rate (Clark et al., 2013; Janiak et al., 2018; Roscito et al., 2018; Wertheim et al., 2015). Reciprocally, when an element becomes relatively more important, its rate is expected to slow. Thus, in cases of phenotypic convergence, rates can be exploited to reveal important genes associated with the phenotype, such as changes to muscle and skin genes associated with the mammalian transition to a marine environment (Chikina et al., 2016) and loss of constraint of vision-related genetic elements in subterranean mammals (Partha et al., 2017; Prudent et al., 2016). Rate shifts can thus provide an evolutionary perspective on the contribution of genes, non-coding elements, or pathways to phenotypes of interest (Hiller et al., 2012). Here we report the genome-wide, pan-mammalian correlations between evolutionary rates of genes and lifespan phenotypes.

In mammals, lifespan is strongly positively correlated with adult body size such that the largest mammals (whales) are longest-lived and the smallest mammals (small rodents) are shortest lived (Figure 1B). However, if lifespan is corrected for body size, species including bats, the naked mole-rat, and some primates are clearly exceptionally long-lived given their body sizes. The term ‘longevity’ is applied to both phenotypes and previous studies have focused on small numbers of species at both phenotypic extremes to find genetic and physiological explanations for the extended phenotypes (Foley et al., 2018; Keane et al., 2015; Kim et al., 2011; Seim et al., 2013; Sulak et al., 2016). In our study we explicitly distinguish two different extended longevity traits: the ‘long-lived large-bodied’ trait (3L) and the ‘exceptionally long-lived given body size’ trait (ELL). Using maximum lifespan and body size data (Tacutu et al., 2018) we define the 3L and ELL phenotypes (Supplementary file 1) to be the first and second principal components of body size and maximum lifespan (Figure 1B, Figure 1—figure supplement 1). The resulting trait values are orthogonal with respect to each other, ensuring that they can be analyzed independently.

Having defined the 3L and ELL phenotypes, we compute the association between these phenotypes and protein-specific relative evolutionary rates (RERs) using the RERconverge method (Kowalczyk et al., 2018; Partha et al., 2018) (Figure 1C and Materials and methods). Relative evolutionary rates quantify the deviation in evolutionary rate of a protein along a specific phylogenetic branch from proteome-wide expectations. Negative RERs indicate fewer substitutions than expected due to increased constraint. Positive RERs correspond to more substitutions than expected, which could arise due to relaxation of constraint or positive selection.

After computing correlations between all protein relative evolutionary rates and the 3L and ELL phenotypes, we find an excess of low p-values (Figure 1D and Figure 1E). In order to evaluate how much of the signal is due to true association with the phenotypes, we used a phylogenetically restricted permutation strategy (termed ‘permulations’, see Materials and methods) to generate null correlation statistics. We find that the real and permuted p-value distributions are indeed different, indicating that a large fraction of genes are correlated with lifespan phenotypes. We quantify the fraction of non-null genes using the π₁ method (Storey, 2003). Using our permulation p-values as the null distribution, the fraction of true positives was inferred to be ~15% (π₁ = 0.153) for the 3L phenotype and ~7.5% (π₁ = 0.075) for the ELL phenotype. Using the theoretical uniform null distribution, the corresponding values are π₁=0.108 and π₁ = 0.021, respectively. Regardless of the null distribution choice, our analysis clearly demonstrates a significant molecular signal for gene evolutionary rates correlated with lifespan phenotypes, with a considerably higher number of associations for the 3L phenotype.

Our analysis investigates both positive and negative correlations between evolutionary rates of genes and changes in lifespan phenotypes. Positive correlations represent genes with faster evolutionary rates in species with high 3L and ELL phenotype values relative to species with low phenotype values. Conversely, negative correlations represent genes with slower evolutionary rates in species with high 3L and ELL phenotype values relative to species with low phenotype values. Note that correlation directionality is relative – faster evolution in low phenotype values corresponds to slower evolution in high phenotype values and vice versa. However, the choice has important consequences for biological interpretation and necessitates an in-depth discussion.

RERs reflect the amount of constraint on the genetic element. A decrease in RER implies greater evolutionary constraint and a greater contribution of that gene to fitness. An increase in RER, on the other hand, has two nearly opposite interpretations. An increased RER could arise because of a relaxation of evolutionary constraint driven by a reduced contribution to fitness. Alternatively, an increased RER could arise due to positive (also termed directional) selection, which implies that that the gene is actively undergoing directed evolution and thus could be contributing to trait-related innovation. However, since there is no default evolutionary rate for protein coding sequence, genes evolving slower in long-lived species could just as easily be interpreted as evolving faster in short-lived species. In our study, we have chosen to interpret the rate changes with reference to the effect in long-lived species (Figure 2). The ancestral mammal is believed to have been small and short-lived, and thus large values of 3L and ELL are derived traits. Consequently, our interpretation reduces to assuming that change in phenotype and shifts in evolutionary rate coincide.

Figure 2

Download asset Open asset

With this interpretation, negative correlations imply that a gene is more important in species with large values of 3L and ELL, while positive correlations imply either a relaxation of constraint or positive selection in species with large values of 3L and ELL. While such positively correlated genes would be of great interest if they indeed represent molecular innovations underlying evolution of extended lifespan, we find relatively fewer such genes, no evidence of positive selection among them (see Supplementary file 6), and no enriched pathways associated with them.

We thus focus our analysis on the negatively correlated genes, which we interpret as being under increased purifying selection in species with high longevity (3L or ELL) values. It is theoretically possible for negative correlations to be caused by accelerated evolution in small and short-lived species because of directional selection associated with development of low longevity values. However, this possibility can be rules out via branch-site models for positive selection using the low longevity species as foreground, as these show little evidence for positive selection (see Supplementary file 6). Together, these analyses support a single interpretation of the main rate convergence signal as a decrease in evolutionary rate, and thus an increase in purifying selection, experienced by species with large 3L and ELL values. It is important to emphasize that these genes are unlikely to have contributed to molecular innovation that lead to the establishment of 3L and ELL traits, but rather these represent existing biological systems that become especially important after the traits are established (see Discussion).

While we observe a clear excess of genes at low p-values, we focus on pathway enrichment analysis which both demonstrates a stronger signal and facilitates interpreting our results in the context of existing knowledge. We investigate enriched pathways for both 3L and ELL phenotypes using a rank-based method (Supplementary file 2). After performing standard multiple-hypothesis testing corrections on the empirical p-values from permulations, there remains considerable pathway-level signal underlying the 3L and ELL traits. Both the gene-level and pathway-level results were highly robust to species removal, which indicates the biological pathways revealed here are important for longevity across mammals and are not restricted to specific species (see Materials and methods, Figure 1—figure supplement 2, Supplementary file 4, and Supplementary file 5).

Among pathways under increased constraint in 3L species, we find a striking abundance of pathways related to cancer control. Those pathways can be organized into the broad categories of ‘cell cycle control’, ‘cell death’, and ‘innate and adaptive immunity’, and they also include other cancer-related pathways such as p53 regulation and telomere maintenance (Figure 3A). We likewise see a significant enrichment in cancer-related genes more broadly. We compared correlation statistics between known cancer genes from Bailey et al. (2018) and found that tumor suppressor genes had significantly lower correlation statistics (rho) than all genes (Wilcoxon rank-sum p-value=4.114e-8) while oncogenes had no significant difference in correlation statistics compared to all genes (Wilcoxon rank-sum p-value=0.3745) (Figure 3—figure supplement 1). This may indicate preferential purifying selection on tumor suppressor genes in large, long-lived species for lower cancer incidence. Considering that these findings, our 3L results can be naturally interpreted in the context of Peto’s Paradox (Peto, 2016). The paradox reasons as follows: if all cells have a similar probability of undergoing a malignant transformation, organisms with more cells should have a greater risk of developing cancer. However, empirical cancer rates do not vary with body size (Peto, 2016), which implies that larger animals harbor mechanisms to suppress cancer rates. Top 3L constrained pathways are associated with multiple cancer control mechanisms, including DNA repair, cell cycle control, cell death, and immune function (Figure 3B and Figure 3C). A normal cell’s transformation to malignancy involves failure of all these processes, and our analysis suggests that 3L animals are invested in the maintenance of each of their associated pathways through increased purifying selection. Based on enrichment and permulation results, we can infer that cell cycle fidelity, an early step in cancer development, is most important over evolutionary time scales for 3L species. Further, there is no evidence for enrichment of pathways associated with metastasis and angiogenesis, later steps in cancer development. This finding suggests that large, long-lived species have experienced increased selective pressure to protect pathways involved in early cancer stages but not later stages, perhaps because the most severely negative fitness impacts of cancer are felt earlier in its development. Species-specific cancer control mechanisms have been identified in individual species, such as increased TP53 copy number in elephants (Sulak et al., 2016), but we show here that investment is cancer control is key to longevity across the entire mammalian phylogeny because top enriched pathways for the 3L phenotype do not depend on a handful of species (Figure 1—figure supplement 3 and Figure 1—figure supplement 2).

Figure 3 with 1 supplement see all

Download asset Open asset

An additional pathway that shows a strong signal of increased constraint with the 3L phenotype is the insulin-like growth factor (IGF1) signaling pathway (Figure 4), which deserves special consideration. Perturbations of IGF1 signaling result in changes in lifespan and body size in diverse organisms (Johnson et al., 2013; Kimura et al., 1997; Stout et al., 2013), which suggests that the IGF1 pathway may be a source of innovation underlying the evolution of the 3L trait. However, we find that IGF1 pathway genes in fact evolve more slowly and are thus under increased purifying selection in large, long-lived species. The magnitude of this signal is quite striking; IGF1 and IGF1R are ranked 1014 and 94 respectively for increased constraint and several other pathway members are near the top (see Figure 3A). A possible explanation is that the IGF1 pathway plays an important role in cancer control (Larsson et al., 2005), and this cancer-related signal of constraint dominates any adaptive signal related to lifespan. There are also further reasons to believe that the IGF1 pathway is not the main source of the 3L trait. Across the mammalian phylogeny, lifespan is strongly correlated with body size, but genetic perturbations in the IGF pathway result in longer-lived individuals that are of smaller size (Holzenberger et al., 2003; Sutter et al., 2007) thus decoupling the two traits. This strongly suggests that changes in the IGF1 pathway are unlikely to drive the natural evolution of the large, long-lived phenotype, which is established through a different, yet-unknown mechanism.

Figure 4

Download asset Open asset

For the ELL phenotype, we find a smaller, more focused set of enriched genes and pathways. Of those significantly enriched constrained pathways for ELL, we see some overlap with functional groups represented in results from the 3L phenotype, notably immune-related and DNA repair pathways (Figure 5A). However, although the functional groups are the same, the pathways contained within them differ between the two phenotypes (File S2, File S3). In particular, the only significantly constrained DNA repair pathways for the 3L phenotype involve Fanconi’s anemia, while the ELL phenotype shows significantly constrained DNA repair pathways for a variety of repair functions (Figure 5). Such pathways stand out not only because of the connection between DNA repair and cancer control, but also because of the observed relationship between DNA repair and aging independent of cancer incidence. This relationship can be demonstrated experimentally by creating double-stranded DNA breaks in laboratory mice to induce an aging phenotype (White et al., 2015). There is also evidence that DNA damage causes dysregulation of the cellular chromatin state and thus can contribute to aging even in post-mitotic cells (Oberdoerffer et al., 2008; Shi and Oberdoerffer, 2012).

Figure 5

Download asset Open asset

In addition to DNA repair-related pathways, we also noted pathways related to NFKB signaling for which overexpression in downstream targets has been associated with aging. Experimental evidence suggests a connection between NFKB signaling and DNA repair through sirtuins, a chromatin regulator family that has already been implicated in lifespan control (Howitz et al., 2003; Mao et al., 2012). Sirtuins mediate DNA damage-induced dysregulation and are also responsible for silencing NFKB-regulated genes, thus connecting the two processes (Salminen et al., 2008). Overall, our analysis strongly suggests that fidelity in DNA repair and NFKB signaling contributes to the fitness of ELL species, indicating that these pathways may be a fruitful avenue for aging research and intervention.

We employed an evolutionary rates-based method as an unbiased, genome-wide, pan-mammalian scan to identify genes and pathways that evolve significantly slower in long-lived large-bodied species (3L) and species that are exceptionally long-lived given their body size (ELL). Pathways related to cancer control, including cell cycle, DNA repair, cell death, and immunity, evolve significantly slower in 3L species, which suggests that cancer resistance is an important functionality the enable to evolution of large and long-lived species. Alternatively, a broader set of DNA repair genes and a more focused set of immune genes related to NFKB signaling evolve significantly slower in ELL species, both of which may be linked to effective DNA repair in order to preserve chromatin state.

Our analysis differs from previous efforts in both methodology and results. Firstly, we do not consider individual amino acid changes within a protein as the unit of convergence, but rather we calculate the overall evolutionary rate of a protein on each branch of a phylogeny and consider these evolutionary rates as the unit of convergence. Secondly, unlike previous studies that focused on lineage- or species-specific changes, we look for correlations between evolutionary rates and quantitative life history traits across the entire phylogeny. This pan-mammalian approach allows us to generalize our findings to describe evolutionary trends throughout all mammals. We also draw a careful distinction between absolute lifespan and relative lifespan, which allows us to make district conclusions about the 3L and the ELL phenotypes.

The most important distinction between our work and previous work is that prior studies began with the assumption that some genes must be under positive selection in association with evolution of extended lifespan. From our unbiased analysis, we find that in fact the strongest signal of evolutionary convergence is that of increased constraint on certain genes and gene families in long-lived species (see Figure 3—figure supplement 1). While some of the pathways have been reported previously (such as cell cycle control, DNA repair, telomerase repair, and IGF1 signaling) our finding is actually the opposite of (though not necessarily contradictory to) the positive selection signal that has been emphasized in previous work, which from the perspective of evolutionary rate is decreased constraint. In fact, we find comparatively few genes and no significant pathway enrichments for the opposite trend – faster relative rates in longer lived species – which would correspond to genes potentially under positive selection in longer-lived species. We therefore focus our discussion on genes and pathways evolving slower in species with large values for the 3L and ELL traits. We considered these genetic elements to be important to the evolution of the 3L and ELL traits because they have been protected from accumulating deleterious mutations and therefore evolve slower.

Despite considerable previous work on genes under positive selection in species with extreme lifespan traits, our analysis did not recover any positively selected pathways in the pan-mammalian analysis. However, for any new trait to arise, some corresponding genetic changes must occur. It is possible that many of the molecular innovations that lead to increased lifespan are species-specific and thus would not be detected by our method. However, it is also true that our relative evolutionary rates method is much more suited to detecting the kind of long-term convergent increases in purifying selection that we observe. Figure 2 represents a simple schematic history of the evolution of a trait and corresponding rate changes. During the establishment of a new trait, some genes experience a brief period of positive selection which generates the molecular innovation to create the new trait, after which the derived sequence will again be subject to purifying selection. Using extant species as a genomic reference for evolutionary history, we can only infer the average rate over the entire history and the brief period of positive selection results in a small, potentially undetectable signal. On the other hand, we can hypothesize another class of genes which become more important as a trait is established and thus experience continuously increased purifying selection after the trait exists. These are the trait ‘enabling’ as opposed to the trait ‘establishing’ genes. Because the rate change in the case of trait-enabling genes is permanent, it has a greater impact on the inferred average rate observable from extant data. Our analyses strongly support that in the case of extended longevity, such ‘enabling’ genes not only exist but are also convergent across independent trait change events.

We have interpreted the rate correlation from the prospective of species with large phenotype-values, thereby assuming phenotype changes and rate changes coincide. The implicit choice of default rate does not affect the final gene ranking on which our analysis is based, but it does have one important consequence. While decreases in rate are easily interpretable as increased constraint, rate increase can be due to relaxation of constraint or positive selection. We address this asymmetry simply by testing both possibilities. We test the relative accelerated genes for positive selection in the long-lived species and the relatively decelerated genes for positive selection in the short-lived species. We find no evidence of systematic positive selection in either direction.

We interpret our strongest pathway enrichment signals as indicating that cancer control is important for enabling evolution of the 3L phenotype, which is in agreement with previous work that has found similar changes in expression levels of cell cycle and immune function genes in both cancerous and aging cells (Chatsirisupachai et al., 2019). However, there may be alternative explanations. Specifically, differential pathogen pressure is expected to affect some of the same pathways. The connection to immune pathways is clear, and viruses are known to co-opt the cell cycle. TP53 and cell death are likewise important mechanisms by which cells control viral infection. All of these pathways may thus be under increased constraint in species more likely to experience viral infections. However, we believe these explanations are unlikely for several reasons. Firstly, we do not see an enrichment of genes that are specialized for virus control, such as MX1, OAS1, DHX58, and genes activated by type I interferon. Secondly, the interaction between pathogen pressure and evolutionary rates has been studied extensively and it is found that it typically drives adaptive changes in pathogen interacting protein as pathogens are often unique to a species (Kosiol et al., 2008; Sackton et al., 2007; Shultz and Sackton, 2019). This is at odds with the increased constraint of immune genes that we observe. Finally, the probability of cancer increases with age and there is theoretical and empirical evidence that importance of cancer resistance increases with body-size and lifespan. On the other hand, for humans the probability of dying from many infections decreases sharply in early life (though increases again at post-reproductive ages) (Palmer et al., 2018) suggesting that infection resistance should not be preferentially selected for in long-lived species.

Overall, the genes and pathways we identified whose constraint is negatively correlated with the 3L and ELL traits encompass functionalities important for the evolution of extended lifespan, and they therefore represent candidate genes and pathways for further experimental exploration. Importantly, such genes were uncovered using an unbiased, genome-wide pan-mammalian scan. As such, these results point to keys to exceptional longevity that are not specific to one or a handful of species, but that are universal across mammals.

RERconverge measures relationships between relative evolutionary rates of orthologous genes and phenotype values based on a set of gene trees whose branch lengths represent protein evolutionary rates and a set of phenotype values, either binary or continuous. The RERconverge package is freely-available for use at https://github.com/nclark-lab/RERconverge along with walkthroughs for beginner users (Kowalczyk et al., 2018). The original RERconverge method was successfully used to find genes undergoing convergent evolutionary rate shifts in marine mammals (Chikina et al., 2016) and subterranean mammals (Partha et al., 2017), and recent statistical improvements have made RERconverge more robust and have given it more power to detect such rate shifts (Partha et al., 2018). Here, the RERconverge methods were extended to use on continuous traits.

To perform the RERconverge analysis, we first used PAML (Yang, 2007) to create maximum likelihood gene trees whose branch lengths represent evolutionary rates by means of number of amino acid substitutions. Gene trees were generated for 19,149 amino acid alignments of orthologous genes from the 62 mammal species from the UCSC 100-way alignment (Blanchette et al., 2004; Harris, 2007; Kent et al., 2002) using the hg19 genome, all of which are available at http://genome.ucsc.edu/. Trees were read into R using the readTrees() function from RERconverge. This step also estimated a master tree with branch lengths that represented the average branch lengths across many gene trees. For all further steps, only 61 species were used as listed in Supplementary file 1; the cape golden mole was excluded because longevity data were not available for it. Relative evolutionary rates corrected for genome-wide evolutionary rates for each species as well as corrected for branch length heteroskedasticity using a weighted regression approach (Partha et al., 2018) were calculated using getAllResiduals() with weight = T, scale = T, cutoff = 0.001, and useSpecies = species names for the 61 mammals species.

Relative evolutionary rates (RERs) quantify the deviation in evolutionary rate of a protein along a specific phylogenetic branch from proteome-wide expectations. Negative RERs indicate fewer substitutions than expected due to increased constraint. Positive RERs correspond to more substitutions than expected, which could arise due to relaxation of constraint or positive selection. To correlate RERs with lifespan phenotypes, we use phenotypic change along phylogenetic branches computed from maximum likelihood ancestral state reconstruction (Revell, 2012). This transformation is equivalent to phylogenetically independent contrasts (Felsenstein, 1985) and thus removes phylogenetic dependence from the phenotype values.

We used the char2Paths() function on each phenotype with trees read in using readTrees() to create a phenotype vector that represented the predicted difference in phenotype values between each species and its ancestor. The char2Paths() function first uses maximum likelihood estimation through the fastAnc() function from phytools (Revell, 2012) and then subtracts the values between connected pairs of species to calculate a phylogenetically-independent measure of the change in the phenotype over evolutionary time. Finally, the getAllCor() function was run with each phenotype and the RER matrix with method = ‘p’, min.pos = 0, and winsorizeRER = 3 to perform Pearson correlations with RERs winsorized such that the two most extreme RER values are set to the value of the third most extreme RER. See Figure 1—figure supplement 4 for evolutionary rate-phenotype scatterplots with high correlation values.

The gene lists produced from the RERconverge correlation analyses were used to calculate pathway enrichments for all canonical pathways from mSigDB (Liberzon et al., 2011) and Mouse Genome Informatics (MGI) functional annotations (Blake et al., 2003; Eppig et al., 2015). For each pathway, a Wilcoxon Rank-Sum statistic was calculated to compare the sign of Rho times the negative log p-value for correlations of genes in the pathway to the same measure for all genes included in a pathway annotation using getStat() from RERconverge to calculate the sign of Rho times the negative log p-value for correlations for each gene and fastwilcoxGMTall() to quickly calculate an approximation of the Wilcoxon Rank-Sum statistic.

Genome alignments are publicly available through the UCSC genome browser - processed alignments in the form of phylogenetic trees that were used for this study are available via the RERconverge package on GitHub. Code for the RERconverge package and vignettes instructing proper use are publicly available on GitHub (https://github.com/nclark-lab/RERconverge; copy archived at https://github.com/elifesciences-publications/RERconverge). Longevity phenotype data is available as supplementary material.

1. 1. Blanchette M
   2. Kent WJ
   3. Riemer C
   4. Elnitski L
   5. Smit AFA
   6. Roskin KM
   7. Baertsch R
   8. Rosenbloom K
   9. Clawson H
   10. Green ED
   11. Haussler D
   12. Miller W

   (2003) UCSC genome browser

   UCSC 100-way alignment.

   http://hgdownload.cse.ucsc.edu/goldenPath/hg19/multiz100way/alignments/knownCanonical.exonAA.fa.gz

1. 1. Bailey MH
   2. Tokheim C
   3. Porta-Pardo E
   4. Sengupta S
   5. Bertrand D
   6. Weerasinghe A
   7. Colaprico A
   8. Wendl MC
   9. Kim J
   10. Reardon B
   11. Kwok-Shing Ng P
   12. Jeong KJ
   13. Cao S
   14. Wang Z
   15. Gao J
   16. Gao Q
   17. Wang F
   18. Liu EM
   19. Mularoni L
   20. Rubio-Perez C
   21. Nagarajan N
   22. Cortés-Ciriano I
   23. Zhou DC
   24. Liang WW
   25. Hess JM
   26. Yellapantula VD
   27. Tamborero D
   28. Gonzalez-Perez A
   29. Suphavilai C
   30. Ko JY
   31. Khurana E
   32. Park PJ
   33. Van Allen EM
   34. Liang H
   35. Lawrence MS
   36. Godzik A
   37. Lopez-Bigas N
   38. Stuart J
   39. Wheeler D
   40. Getz G
   41. Chen K
   42. Lazar AJ
   43. Mills GB
   44. Karchin R
   45. Ding L
   46. MC3 Working Group, Cancer Genome Atlas Research Network

   (2018) Comprehensive characterization of Cancer driver genes and mutations

   Cell 174:1034–1035.

   https://doi.org/10.1016/j.cell.2018.07.034

   • PubMed
   • Google Scholar
2. 1. Bininda-Emonds OR
   2. Cardillo M
   3. Jones KE
   4. MacPhee RD
   5. Beck RM
   6. Grenyer R
   7. Price SA
   8. Vos RA
   9. Gittleman JL
   10. Purvis A

   (2007) The delayed rise of present-day mammals

   Nature 446:507–512.

   https://doi.org/10.1038/nature05634

   • PubMed
   • Google Scholar
3. 1. Blake JA
   2. Richardson JE
   3. Bult CJ
   4. Kadin JA
   5. Eppig JT
   6. Mouse Genome Database Group

   (2003) MGD: the mouse genome database

   Nucleic Acids Research 31:193–195.

   https://doi.org/10.1093/nar/gkg047

   • PubMed
   • Google Scholar
4. 1. Blanchette M
   2. Kent WJ
   3. Riemer C
   4. Elnitski L
   5. Smit AF
   6. Roskin KM
   7. Baertsch R
   8. Rosenbloom K
   9. Clawson H
   10. Green ED
   11. Haussler D
   12. Miller W

   (2004) Aligning multiple genomic sequences with the threaded blockset aligner

   Genome Research 14:708–715.

   https://doi.org/10.1101/gr.1933104

   • PubMed
   • Google Scholar
5. 1. Chatsirisupachai K
   2. Palmer D
   3. Ferreira S
   4. de Magalhães JP

   (2019) A human tissue-specific transcriptomic analysis reveals a complex relationship between aging, Cancer, and cellular senescence

   Aging Cell 18:e13041.

   https://doi.org/10.1111/acel.13041

   • PubMed
   • Google Scholar
6. 1. Chikina M
   2. Robinson JD
   3. Clark NL

   (2016) Hundreds of genes experienced convergent shifts in selective pressure in marine mammals

   Molecular Biology and Evolution 33:2182–2192.

   https://doi.org/10.1093/molbev/msw112

   • PubMed
   • Google Scholar
7. 1. Clark NL
   2. Alani E
   3. Aquadro CF

   (2013) Evolutionary rate covariation in meiotic proteins results from fluctuating evolutionary pressure in yeasts and mammals

   Genetics 193:529–538.

   https://doi.org/10.1534/genetics.112.145979

   • PubMed
   • Google Scholar
8. 1. Eppig JT
   2. Blake JA
   3. Bult CJ
   4. Kadin JA
   5. Richardson JE
   6. Mouse Genome Database Group

   (2015) The mouse genome database (MGD): facilitating mouse as a model for human biology and disease

   Nucleic Acids Research 43:D726–D736.

   https://doi.org/10.1093/nar/gku967

   • PubMed
   • Google Scholar
9. 1. Felsenstein J

   (1985) Phylogenies and the comparative method

   The American Naturalist 125:1–15.

   https://doi.org/10.1086/284325

   • Google Scholar
10. 1. Foley NM
    2. Hughes GM
    3. Huang Z
    4. Clarke M
    5. Jebb D
    6. Whelan CV
    7. Petit EJ
    8. Touzalin F
    9. Farcy O
    10. Jones G
    11. Ransome RD
    12. Kacprzyk J
    13. O'Connell MJ
    14. Kerth G
    15. Rebelo H
    16. Rodrigues L
    17. Puechmaille SJ
    18. Teeling EC

    (2018) Growing old, yet staying young: the role of telomeres in bats' exceptional longevity

    Science Advances 4:eaao0926.

    https://doi.org/10.1126/sciadv.aao0926

    • PubMed
    • Google Scholar
11. 1. Harmon LJ
    2. Weir JT
    3. Brock CD
    4. Glor RE
    5. Challenger W

    (2008) GEIGER: investigating evolutionary radiations

    Bioinformatics 24:129–131.

    https://doi.org/10.1093/bioinformatics/btm538

    • Google Scholar
12. Thesis

    1. Harris RS

    (2007)

    Improved Pairwise Alignment of Genomic DNA

    Pennsylvania State University.

    • Google Scholar
13. 1. Hiller M
    2. Schaar BT
    3. Indjeian VB
    4. Kingsley DM
    5. Hagey LR
    6. Bejerano G

    (2012) A "forward genomics" approach links genotype to phenotype using independent phenotypic losses among related species

    Cell Reports 2:817–823.

    https://doi.org/10.1016/j.celrep.2012.08.032

    • PubMed
    • Google Scholar
14. 1. Holzenberger M
    2. Dupont J
    3. Ducos B
    4. Leneuve P
    5. Géloën A
    6. Even PC
    7. Cervera P
    8. Le Bouc Y

    (2003) IGF-1 receptor regulates lifespan and resistance to oxidative stress in mice

    Nature 421:182–187.

    https://doi.org/10.1038/nature01298

    • PubMed
    • Google Scholar
15. 1. Howitz KT
    2. Bitterman KJ
    3. Cohen HY
    4. Lamming DW
    5. Lavu S
    6. Wood JG
    7. Zipkin RE
    8. Chung P
    9. Kisielewski A
    10. Zhang LL
    11. Scherer B
    12. Sinclair DA

    (2003) Small molecule activators of sirtuins extend Saccharomyces cerevisiae lifespan

    Nature 425:191–196.

    https://doi.org/10.1038/nature01960

    • PubMed
    • Google Scholar
16. 1. Janiak MC
    2. Chaney ME
    3. Tosi AJ

    (2018) Evolution of acidic mammalian chitinase genes (CHIA) Is related to body mass and insectivory in primates

    Molecular Biology and Evolution 35:607–622.

    https://doi.org/10.1093/molbev/msx312

    • PubMed
    • Google Scholar
17. 1. Johnson SC
    2. Rabinovitch PS
    3. Kaeberlein M

    (2013) mTOR is a key modulator of ageing and age-related disease

    Nature 493:338–345.

    https://doi.org/10.1038/nature11861

    • PubMed
    • Google Scholar
18. 1. Jones OR
    2. Scheuerlein A
    3. Salguero-Gómez R
    4. Camarda CG
    5. Schaible R
    6. Casper BB
    7. Dahlgren JP
    8. Ehrlén J
    9. García MB
    10. Menges ES
    11. Quintana-Ascencio PF
    12. Caswell H
    13. Baudisch A
    14. Vaupel JW

    (2014) Diversity of ageing across the tree of life

    Nature 505:169–173.

    https://doi.org/10.1038/nature12789

    • PubMed
    • Google Scholar
19. 1. Keane M
    2. Semeiks J
    3. Webb AE
    4. Li YI
    5. Quesada V
    6. Craig T
    7. Madsen LB
    8. van Dam S
    9. Brawand D
    10. Marques PI
    11. Michalak P
    12. Kang L
    13. Bhak J
    14. Yim H-S
    15. Grishin NV
    16. Nielsen NH
    17. Heide-Jørgensen MP
    18. Oziolor EM
    19. Matson CW
    20. Church GM
    21. Stuart GW
    22. Patton JC
    23. George JC
    24. Suydam R
    25. Larsen K
    26. López-Otín C
    27. O’Connell MJ
    28. Bickham JW
    29. Thomsen B
    30. de Magalhães JP

    (2015) Insights into the evolution of longevity from the bowhead whale genome

    Cell Reports 10:112–122.

    https://doi.org/10.1016/j.celrep.2014.12.008

    • Google Scholar
20. 1. Kent WJ
    2. Sugnet CW
    3. Furey TS
    4. Roskin KM
    5. Pringle TH
    6. Zahler AM
    7. Haussler D

    (2002) The human genome browser at UCSC

    Genome Research 12:996–1006.

    https://doi.org/10.1101/gr.229102

    • PubMed
    • Google Scholar
21. 1. Kim EB
    2. Fang X
    3. Fushan AA
    4. Huang Z
    5. Lobanov AV
    6. Han L
    7. Marino SM
    8. Sun X
    9. Turanov AA
    10. Yang P
    11. Yim SH
    12. Zhao X
    13. Kasaikina MV
    14. Stoletzki N
    15. Peng C
    16. Polak P
    17. Xiong Z
    18. Kiezun A
    19. Zhu Y
    20. Chen Y
    21. Kryukov GV
    22. Zhang Q
    23. Peshkin L
    24. Yang L
    25. Bronson RT
    26. Buffenstein R
    27. Wang B
    28. Han C
    29. Li Q
    30. Chen L
    31. Zhao W
    32. Sunyaev SR
    33. Park TJ
    34. Zhang G
    35. Wang J
    36. Gladyshev VN

    (2011) Genome sequencing reveals insights into physiology and longevity of the naked mole rat

    Nature 479:223–227.

    https://doi.org/10.1038/nature10533

    • PubMed
    • Google Scholar
22. 1. Kimura KD
    2. Tissenbaum HA
    3. Liu Y
    4. Ruvkun G

    (1997) daf-2, an insulin receptor-like gene that regulates longevity and diapause in Caenorhabditis elegans

    Science 277:942–946.

    https://doi.org/10.1126/science.277.5328.942

    • PubMed
    • Google Scholar
23. 1. Kosiol C
    2. Vinar T
    3. da Fonseca RR
    4. Hubisz MJ
    5. Bustamante CD
    6. Nielsen R
    7. Siepel A

    (2008) Patterns of positive selection in six mammalian genomes

    PLOS Genetics 4:e1000144.

    https://doi.org/10.1371/journal.pgen.1000144

    • PubMed
    • Google Scholar
24. 1. Kowalczyk A
    2. Meyer WK
    3. Partha R
    4. Mao W
    5. Clark NL
    6. Chikina M

    (2018) RERconverge: an R package for associating evolutionary rates with convergent traits

    Bioinformatics 35:4815–4817.

    https://doi.org/10.1093/bioinformatics/btz468

    • Google Scholar
25. Software

    1. Kowalczyk A
    2. Partha R
    3. Clark NL
    4. Chikina M

    (2019) RERconverge, version c1a9883

    GitHub.

    https://github.com/nclark-lab/RERconverge
26. 1. Larsson O
    2. Girnita A
    3. Girnita L

    (2005) Role of insulin-like growth factor 1 receptor signalling in Cancer

    British Journal of Cancer 92:2097–2101.

    https://doi.org/10.1038/sj.bjc.6602627

    • PubMed
    • Google Scholar
27. 1. Liberzon A
    2. Subramanian A
    3. Pinchback R
    4. Thorvaldsdóttir H
    5. Tamayo P
    6. Mesirov JP

    (2011) Molecular signatures database (MSigDB) 3.0

    Bioinformatics 27:1739–1740.

    https://doi.org/10.1093/bioinformatics/btr260

    • PubMed
    • Google Scholar
28. 1. Lodato MA
    2. Rodin RE
    3. Bohrson CL
    4. Coulter ME
    5. Barton AR
    6. Kwon M
    7. Sherman MA
    8. Vitzthum CM
    9. Luquette LJ
    10. Yandava CN
    11. Yang P
    12. Chittenden TW
    13. Hatem NE
    14. Ryu SC
    15. Woodworth MB
    16. Park PJ
    17. Walsh CA

    (2018) Aging and neurodegeneration are associated with increased mutations in single human neurons

    Science 359:555–559.

    https://doi.org/10.1126/science.aao4426

    • PubMed
    • Google Scholar
29. 1. Mao Z
    2. Tian X
    3. Van Meter M
    4. Ke Z
    5. Gorbunova V
    6. Seluanov A

    (2012) Sirtuin 6 (SIRT6) rescues the decline of homologous recombination repair during replicative senescence

    PNAS 109:11800–11805.

    https://doi.org/10.1073/pnas.1200583109

    • PubMed
    • Google Scholar
30. Book

    1. Medawar PB

    (1952)

    An Unsolved Problem in Biology

    Lewis HK, editors. London: Scientific Research.

    • Google Scholar
31. 1. Meredith RW
    2. Janečka JE
    3. Gatesy J
    4. Ryder OA
    5. Fisher CA
    6. Teeling EC
    7. Goodbla A
    8. Eizirik E
    9. Simão TL
    10. Stadler T
    11. Rabosky DL
    12. Honeycutt RL
    13. Flynn JJ
    14. Ingram CM
    15. Steiner C
    16. Williams TL
    17. Robinson TJ
    18. Burk-Herrick A
    19. Westerman M
    20. Ayoub NA
    21. Springer MS
    22. Murphy WJ

    (2011) Impacts of the cretaceous terrestrial revolution and KPg extinction on mammal diversification

    Science 334:521–524.

    https://doi.org/10.1126/science.1211028

    • PubMed
    • Google Scholar
32. 1. Meyer WK
    2. Jamison J
    3. Richter R
    4. Woods SE
    5. Partha R
    6. Kowalczyk A
    7. Kronk C
    8. Chikina M
    9. Bonde RK
    10. Crocker DE
    11. Gaspard J
    12. Lanyon JM
    13. Marsillach J
    14. Furlong CE
    15. Clark NL

    (2018) Ancient convergent losses of paraoxonase 1 yield potential risks for modern marine mammals

    Science 361:591–594.

    https://doi.org/10.1126/science.aap7714

    • PubMed
    • Google Scholar
33. 1. Muntané G
    2. Farré X
    3. Rodríguez JA
    4. Pegueroles C
    5. Hughes DA
    6. de Magalhães JP
    7. Gabaldón T
    8. Navarro A

    (2018) Biological processes modulating longevity across primates: a phylogenetic Genome-Phenome analysis

    Molecular Biology and Evolution 35:1990–2004.

    https://doi.org/10.1093/molbev/msy105

    • PubMed
    • Google Scholar
34. Book

    1. Nowak RM

    (1999)

    Walker’s Mammals of the World (6th ed)

    Baltimore: The Johns Hopkins University Press.

    • Google Scholar
35. 1. Oberdoerffer P
    2. Michan S
    3. McVay M
    4. Mostoslavsky R
    5. Vann J
    6. Park SK
    7. Hartlerode A
    8. Stegmuller J
    9. Hafner A
    10. Loerch P
    11. Wright SM
    12. Mills KD
    13. Bonni A
    14. Yankner BA
    15. Scully R
    16. Prolla TA
    17. Alt FW
    18. Sinclair DA

    (2008) SIRT1 redistribution on chromatin promotes genomic stability but alters gene expression during aging

    Cell 135:907–918.

    https://doi.org/10.1016/j.cell.2008.10.025

    • PubMed
    • Google Scholar
36. 1. Palmer S
    2. Albergante L
    3. Blackburn CC
    4. Newman TJ

    (2018) Thymic involution and rising disease incidence with age

    PNAS 115:1883–1888.

    https://doi.org/10.1073/pnas.1714478115

    • PubMed
    • Google Scholar
37. 1. Partha R
    2. Chauhan BK
    3. Ferreira Z
    4. Robinson JD
    5. Lathrop K
    6. Nischal KK
    7. Chikina M
    8. Clark NL

    (2017) Subterranean mammals show convergent regression in ocular genes and enhancers, along with adaptation to tunneling

    eLife 6:e25884.

    https://doi.org/10.7554/eLife.25884

    • PubMed
    • Google Scholar
38. Preprint

    1. Partha R
    2. Kowalczyk A
    3. Clark NL
    4. Chikina M

    (2018) Robust methods for detecting convergent shifts in evolutionary rates

    bioRxiv.

    https://doi.org/10.1101/457309

    • Google Scholar
39. 1. Peto R

    (2016) Epidemiology, multistage models, and short-term mutagenicity tests

    International Journal of Epidemiology 45:621–637.

    https://doi.org/10.1093/ije/dyv199

    • PubMed
    • Google Scholar
40. 1. Prudent X
    2. Parra G
    3. Schwede P
    4. Roscito JG
    5. Hiller M

    (2016) Controlling for phylogenetic relatedness and evolutionary rates improves the discovery of associations between species' Phenotypic and genomic differences

    Molecular Biology and Evolution 33:2135–2150.

    https://doi.org/10.1093/molbev/msw098

    • PubMed
    • Google Scholar
41. 1. Revell LJ

    (2012) Phytools: an R package for phylogenetic comparative biology (and other things)

    Methods in Ecology and Evolution 3:217–223.

    https://doi.org/10.1111/j.2041-210X.2011.00169.x

    • Google Scholar
42. 1. Rodríguez JA
    2. Marigorta UM
    3. Hughes DA
    4. Spataro N
    5. Bosch E
    6. Navarro A

    (2017) Antagonistic pleiotropy and mutation accumulation influence human senescence and disease

    Nature Ecology & Evolution 1:0055.

    https://doi.org/10.1038/s41559-016-0055

    • Google Scholar
43. 1. Roscito JG
    2. Sameith K
    3. Parra G
    4. Langer BE
    5. Petzold A
    6. Moebius C
    7. Bickle M
    8. Rodrigues MT
    9. Hiller M

    (2018) Phenotype loss is associated with widespread divergence of the gene regulatory landscape in evolution

    Nature Communications 9:07122.

    https://doi.org/10.1038/s41467-018-07122-z

    • Google Scholar
44. 1. Sackton TB
    2. Lazzaro BP
    3. Schlenke TA
    4. Evans JD
    5. Hultmark D
    6. Clark AG

    (2007) Dynamic evolution of the innate immune system in Drosophila

    Nature Genetics 39:1461–1468.

    https://doi.org/10.1038/ng.2007.60

    • PubMed
    • Google Scholar
45. 1. Salminen A
    2. Huuskonen J
    3. Ojala J
    4. Kauppinen A
    5. Kaarniranta K
    6. Suuronen T

    (2008) Activation of innate immunity system during aging: nf-kb signaling is the molecular culprit of inflamm-aging

    Ageing Research Reviews 7:83–105.

    https://doi.org/10.1016/j.arr.2007.09.002

    • PubMed
    • Google Scholar
46. 1. Seim I
    2. Fang X
    3. Xiong Z
    4. Lobanov AV
    5. Huang Z
    6. Ma S
    7. Feng Y
    8. Turanov AA
    9. Zhu Y
    10. Lenz TL
    11. Gerashchenko MV
    12. Fan D
    13. Hee Yim S
    14. Yao X
    15. Jordan D
    16. Xiong Y
    17. Ma Y
    18. Lyapunov AN
    19. Chen G
    20. Kulakova OI
    21. Sun Y
    22. Lee SG
    23. Bronson RT
    24. Moskalev AA
    25. Sunyaev SR
    26. Zhang G
    27. Krogh A
    28. Wang J
    29. Gladyshev VN

    (2013) Genome analysis reveals insights into physiology and longevity of the Brandt's bat Myotis brandtii

    Nature Communications 4:2212.

    https://doi.org/10.1038/ncomms3212

    • PubMed
    • Google Scholar
47. 1. Shi L
    2. Oberdoerffer P

    (2012) Chromatin dynamics in DNA double-strand break repair

    Biochimica Et Biophysica Acta (BBA) - Gene Regulatory Mechanisms 1819:811–819.

    https://doi.org/10.1016/j.bbagrm.2012.01.002

    • Google Scholar
48. 1. Shultz AJ
    2. Sackton TB

    (2019) Immune genes are hotspots of shared positive selection across birds and mammals

    eLife 8:e41815.

    https://doi.org/10.7554/eLife.41815

    • PubMed
    • Google Scholar
49. 1. Storey JD

    (2003) The positive false discovery rate: a bayesian interpretation and the q -value

    The Annals of Statistics 31:2013–2035.

    https://doi.org/10.1214/aos/1074290335

    • Google Scholar
50. 1. Stout GJ
    2. Stigter EC
    3. Essers PB
    4. Mulder KW
    5. Kolkman A
    6. Snijders DS
    7. van den Broek NJ
    8. Betist MC
    9. Korswagen HC
    10. Macinnes AW
    11. Brenkman AB

    (2013) Insulin/IGF-1-mediated longevity is marked by reduced protein metabolism

    Molecular Systems Biology 9:679.

    https://doi.org/10.1038/msb.2013.35

    • PubMed
    • Google Scholar
51. 1. Subramanian A
    2. Tamayo P
    3. Mootha VK
    4. Mukherjee S
    5. Ebert BL
    6. Gillette MA
    7. Paulovich A
    8. Pomeroy SL
    9. Golub TR
    10. Lander ES
    11. Mesirov JP

    (2005) Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles

    PNAS 102:15545–15550.

    https://doi.org/10.1073/pnas.0506580102

    • PubMed
    • Google Scholar
52. 1. Sulak M
    2. Fong L
    3. Mika K
    4. Chigurupati S
    5. Yon L
    6. Mongan NP
    7. Emes RD
    8. Lynch VJ

    (2016) TP53 copy number expansion is associated with the evolution of increased body size and an enhanced DNA damage response in elephants

    eLife 5:e11994.

    https://doi.org/10.7554/eLife.11994

    • PubMed
    • Google Scholar
53. 1. Sutter NB
    2. Bustamante CD
    3. Chase K
    4. Gray MM
    5. Zhao K
    6. Zhu L
    7. Padhukasahasram B
    8. Karlins E
    9. Davis S
    10. Jones PG
    11. Quignon P
    12. Johnson GS
    13. Parker HG
    14. Fretwell N
    15. Mosher DS
    16. Lawler DF
    17. Satyaraj E
    18. Nordborg M
    19. Lark KG
    20. Wayne RK
    21. Ostrander EA

    (2007) A single IGF1 allele is a major determinant of small size in dogs

    Science 316:112–115.

    https://doi.org/10.1126/science.1137045

    • PubMed
    • Google Scholar
54. 1. Tacutu R
    2. Thornton D
    3. Johnson E
    4. Budovsky A
    5. Barardo D
    6. Craig T
    7. Diana E
    8. Lehmann G
    9. Toren D
    10. Wang J
    11. Fraifeld VE
    12. de Magalhães JP

    (2018) Human ageing genomic resources: new and updated databases

    Nucleic Acids Research 46:D1083–D1090.

    https://doi.org/10.1093/nar/gkx1042

    • PubMed
    • Google Scholar
55. 1. Wertheim JO
    2. Murrell B
    3. Smith MD
    4. Kosakovsky Pond SL
    5. Scheffler K

    (2015) RELAX: detecting relaxed selection in a phylogenetic framework

    Molecular Biology and Evolution 32:820–832.

    https://doi.org/10.1093/molbev/msu400

    • PubMed
    • Google Scholar
56. 1. White RR
    2. Milholland B
    3. de Bruin A
    4. Curran S
    5. Laberge RM
    6. van Steeg H
    7. Campisi J
    8. Maslov AY
    9. Vijg J

    (2015) Controlled induction of DNA double-strand breaks in the mouse liver induces features of tissue ageing

    Nature Communications 6:6790.

    https://doi.org/10.1038/ncomms7790

    • PubMed
    • Google Scholar
57. 1. Williams GC

    (1957) Pleiotropy, natural selection, and the evolution of senescence

    Evolution 11:398.

    https://doi.org/10.2307/2406060

    • Google Scholar
58. 1. Yang Z

    (2007) PAML 4: phylogenetic analysis by maximum likelihood

    Molecular Biology and Evolution 24:1586–1591.

    https://doi.org/10.1093/molbev/msm088

    • PubMed
    • Google Scholar
59. 1. Zhang J
    2. Yang JR

    (2015) Determinants of the rate of protein sequence evolution

    Nature Reviews Genetics 16:409–420.

    https://doi.org/10.1038/nrg3950

    • PubMed
    • Google Scholar

Author details

1. Amanda Kowalczyk

   1. Joint Carnegie Mellon University-University of Pittsburgh PhD Program in Computational Biology, Pittsburgh, United States
   2. Department of Computational and Systems Biology, University of Pittsburgh, Pittsburgh, United States

   Contribution

   Conceptualization, Data curation, Software, Formal analysis, Investigation, Visualization, Methodology

   For correspondence

   kowaae22@pitt.edu

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-9061-1336

2. Raghavendran Partha

   1. Joint Carnegie Mellon University-University of Pittsburgh PhD Program in Computational Biology, Pittsburgh, United States
   2. Department of Computational and Systems Biology, University of Pittsburgh, Pittsburgh, United States

   Contribution

   Conceptualization, Resources, Data curation, Software, Formal analysis, Supervision, Funding acquisition, Visualization, Methodology, Project administration

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-7900-4375

3. Nathan L Clark

   1. Department of Computational and Systems Biology, University of Pittsburgh, Pittsburgh, United States
   2. Pittsburgh Center for Evolutionary Biology and Medicine, University of Pittsburgh, Pittsburgh, United States
   3. Department of Human Genetics, University of Utah, Salt Lake City, United States

   Contribution

   Conceptualization, Resources, Supervision, Funding acquisition, Project administration

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-0006-8374

4. Maria Chikina

   Department of Computational and Systems Biology, University of Pittsburgh, Pittsburgh, United States

   Contribution

   Conceptualization, Resources, Data curation, Software, Supervision, Funding acquisition, Visualization, Methodology, Project administration

   For correspondence

   mchikina@pitt.edu

   Competing interests

   No competing interests declared

• 5,080

  views

• 587

  downloads

• 67

  citations

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