Bayesian Detection of Convergent Rate Changes of Conserved Noncoding Elements on Phylogenetic Trees

May combine two claims.

TraitRate only estimates the likelihood ratio indicating the association between sequence evolution and a given trait; it does not model the pattern of shifts in substitution rate explicitly.

“Additionally, TraitRate only estimates the likelihood ratio indicating the association between sequence evolution and a given trait; it does not model the pattern of shifts in substitution rate explicitly.”

• cites Mayrose and Otto 2011 sourceDisputed

  The Introduction (page not provided) describes an explicit model in which evolutionary rates depend on a lineage’s trait state over time. The Combined Character and Sequence Model and Likelihood Computations sections (pages not provided) specify a relative substitution-rate parameter and model rate shifts along branches. The paper also uses a likelihood-ratio test, but does not only estimate that ratio. (LLM, conf=0.90)

  Checked by a reasoning model — may be wrong.

  Experimental — accuracy not yet measured
• cites Levy Karin et al. 2017 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

The Forward Genomics method tests the significance of the Pearson correlation between normalized substitutions and hypothetical phenotypic states on each branch.

“The “Forward Genomics” method (Hiller et al. 2012; Prudent et al. 2016) tests the significance of the Pearson correlation between normalized substitutions and hypothetical phenotypic states on each branch.”

• cites Hiller et al. 2012 sourceDisputed

  The Experimental Procedures section (“Screening for conserved regions that evolve faster in species lacking a trait”; page unavailable) describes a different test: it compares percent identity between trait-loss and trait-preserving species and counts species that violate a 1% separation. It does not describe testing a Pearson correlation between branch-level normalized substitutions and hypothetical phenotypic states. (LLM, conf=0.88)

  Checked by a reasoning model — may be wrong.

  Experimental — accuracy not yet measured
• cites Prudent et al. 2016 sourceSupported

May combine two claims.

Marcovitz et al. (2016) used parsimony to reconstruct traits and genome transitions and estimate their correlations, but were unable to account for the uncertainty of conservation pattern estimated from the sequencing data or the probability of a chance match between sequence evolutionary history and phenotype.

“For example, Marcovitz et al. (2016) used parsimony to reconstruct traits and genome transitions and estimate their correlations, but in doing so were unable to account for the uncertainty of conservation pattern estimated from the sequencing data or the probability of a chance match between sequence evolutionary history and phenotype.”

• cites Marcovitz et al. 2016 sourceDisputed

  Split into 2 atoms — 1 confirmed, 0 partially supported, 1 disputed, 0 misattributed, 0 unverifiable. Expand the claim to see per-atom verdicts.

  Checked by a reasoning model — may be wrong.

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S2 (checkable).

We generally do not have a robust understanding of the genomic changes underlying phenotypic convergence.

“However, we generally do not have a robust understanding of the genomic changes underlying phenotypic convergence”

• cites Wray 2013 sourceNot checked — no full text of the source
• cites Rosenblum et al. 2014 sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Booker et al. discovered 166 bat-accelerated regions overlapping with enhancers in developing mouse limbs, including one that likely regulated expression of the HoxD cluster important for forelimb development.

“Booker et al. (2016) discovered 166 bat-accelerated regions overlapping with enhancers in developing mouse limbs, including one that likely regulated expression of the HoxD cluster important for forelimb development.”

• cites Booker et al. (2016) sourcePartly supported

  The Abstract reports 166 bat-accelerated regions overlapping H3K27ac and p300 peaks in developing mouse limbs, but does not establish that all 166 overlap confirmed enhancers. It also reports that BAR116 has forelimb enhancer activity and could be involved in regulating the bat HoxD complex; “likely regulated” is stronger than that conclusion. The indexed passages provide section labels but no page numbers. (LLM, conf=0.84)

  Experimental — accuracy not yet measured

Classified as probably not a substantial claim.

Phenotypic convergence is generally believed to be a strong signature of natural selection.

“Phenotypic convergence, in which the same function or morphology evolves multiple times independently, often due to adaption to similar environmental changes, is generally believed to be a strong signature of natural selection”

• cites Kishida et al. 2007 sourceCould not verify

  The paper’s introduction (p. 428) and results discuss olfactory-receptor gene loss in marine vertebrates. Searches for phenotypic convergence and natural selection found no passage in our copy stating that convergence is generally considered a strong signature of natural selection. (LLM, conf=0.72)

  Experimental — accuracy not yet measured
• cites Brawand et al. 2008 sourceCould not verify

  The Results/Discussion section on functional and disabled monotreme VIT genes reports purifying selection on a platypus gene, not phenotypic convergence. Searches for phenotypic convergence and independently evolved adaptations found no passage in our copy saying that convergence is generally a strong signature of natural selection. (LLM, conf=0.76)

  Experimental — accuracy not yet measured
• cites Stern 2013 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Holloway et al. identified 4,797 regions accelerated at the base of therian mammals, many of which are noncoding and located close to developmental transcription factors.

“Holloway et al. (2016) identified 4,797 regions accelerated at the base of therian mammals, many of which are noncoding and located close to developmental transcription factors.”

• cites Holloway et al. (2016) sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing).

Kostka et al. discovered similar lineage-specific accelerated regions near developmental genes in all five apes.

“Kostka et al. (2018) discovered similar lineage-specific accelerated regions near developmental genes in all five apes.”

• cites Kostka et al. (2018) sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing).

Numerous studies have used changes in sequence conservation of conserved elements to identify regulatory regions that may be important for lineage-specific phenotypes.

“Numerous studies have used changes in sequence conservation of conserved elements as means to identify regulatory regions which may be of particular importance for lineage-specific phenotypes.”

• cites Pollard et al. (2006) sourceSupported
• cites Kostka et al. (2018) sourceSupported
• cites Holloway et al. (2016) sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing).

Convergence at the molecular level can arise because of identical substitutions.

“Convergence at the molecular level can arise because of identical substitutions, and convergent rate changes can arise via consistent shifts in substitution rate in genomic regions influencing particular phenotypic targets of natural selection”

• cites Chikina et al. 2016 sourceNot checked — no full text of the source
• cites Partha et al. 2017 sourceSupported
• cites Muntané et al. 2018 sourceSupported

May combine two claims.

Coevol jointly models parameters of substitution process and continuous phenotypes as a multivariate Brownian process and outputs the posterior distribution of the correlation between phenotypes and substitution parameters, but does not allow for discrete phenotypes.

“Coevol (Lartillot and Poujol 2011) jointly models parameters of substitution process (e.g., substitution rates) and continuous phenotypes as a multivariate Brownian process and outputs the posterior distribution of the correlation between phenotypes and substitution parameters, but does not allow for discrete phenotypes.”

• cites Lartillot and Poujol 2011 sourceSupported

Classified as probably not a substantial claim. May combine two claims.

The model permits independent gain and loss of conservation on multiple lineages and encourages nearby branches to have the same state and substitution rate, which is reasonable for closely related species and branches in a phylogeny.

“Such a model permits independent gain and loss of conservation on multiple lineages and also encourages nearby branches to have the same state and substitution rate, which is reasonable for closely related species and branches in a phylogeny and is also a common assumption in phylogenetics”

• cites Thorne et al. 1998 sourcePartly supported

  The Hierarchical Model of Rate Evolution section allows rates to differ among branches and models autocorrelation between ancestral and descendant branch rates. It does not model conservation states or independent gains and losses of conservation. The Discussion and Conclusions section also says widely used phylogeny-reconstruction methods ignore the potential information from rate autocorrelation, rather than establishing it as a common assumption. (LLM, conf=0.88)

  Experimental — accuracy not yet measured
• cites Rannala and Yang 2007 sourceNot checked — no full text of the source

• AssumptionThe authors state this as an assumption: The substitution rate of a given branch is expected to correlate with that of its parent branch.
• AssumptionThe authors state this as an assumption: For each element, a limited number of rate categories occur on the phylogeny.

May combine two claims.

TraitRate models the probability of rate shifts along a fixed ultrametric tree, an approach that was shown to deteriorate the performance in practice.

“This method models the probability of rate shifts along a fixed ultrametric tree, an approach that was shown to deteriorate the performance in practice (Mayrose and Otto 2011).”

• cites Mayrose and Otto 2011 sourcePartly supported

  Likelihood Computations describes mapping character histories onto an input clocklike tree, while also allowing calculations across multiple plausible ultrametric trees; the tree need not be fixed. Phylogenetic Uncertainty reports reduced power in simulations when inferred gene trees were inaccurately ultrametricized. It does not show that using a fixed ultrametric tree generally deteriorates performance in practice. (LLM, conf=0.86)

  Experimental — accuracy not yet measured

Classified as probably not a substantial claim.

Several existing relaxed clock models explicitly model substitution rate variation across phylogeny.

“Several existing relaxed clock models (Drummond and Suchard 2010; Heath et al. 2012) explicitly model substitution rate variation across phylogeny.”

• cites Drummond and Suchard 2010 sourceSupported
• cites Heath et al. 2012 sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Pollard et al. identified 202 regions accelerated in the human genome but conserved in other vertebrates, some of which are RNA genes and tissue-specific enhancers.

“For example, Pollard et al. (2006) identified 202 regions accelerated in the human genome but conserved in other vertebrates, some of which are RNA genes and tissue-specific enhancers.”

• cites Pollard et al. (2006) sourcePartly supported

  The Results section, “Human Acceleration in Predicted Functional Elements” (page unavailable), reports 202 regions conserved across vertebrates but accelerated in humans. The Discussion identifies HAR1 as a structural RNA expressed during neocortical development. The paper suggests possible transcriptional regulatory roles for other regions but does not establish that any of the 202 are tissue-specific enhancers. (LLM, conf=0.87)

  Experimental — accuracy not yet measured

Classified as probably not a substantial claim. May combine two claims.

Recent phylogenetic work supports the conclusion that the ratites are paraphyletic, implying convergent loss of flight in these lineages.

“Recent phylogenetic work supports the conclusion that the ratites (including ostrich, emu, cassowaries, kiwis, rheas, and the extinct moas and elephant bird) are paraphyletic, implying convergent loss of flight in these lineages”

• cites Harshman et al. 2008 sourcePartly supported

  The Results and Discussion place flighted tinamous within the flightless ratites and infer at least three independent losses of flight, supporting the claim’s central evolutionary point. The paper explicitly calls ratites “polyphyletic,” not “paraphyletic,” and its stated minimum of three losses concerns ostriches, rheas, and Australasian ratites rather than establishing separate losses for the extinct moas and elephant birds. (LLM, conf=0.88)

  Experimental — accuracy not yet measured
• cites Baker et al. 2014 sourceNot checked — no full text of the source
• cites Mitchell et al. 2014 sourceNot checked — no full text of the source

Classified as probably not a substantial claim.

Regions of the genome outside protein-coding genes exhibit considerable levels of conservation across evolutionary time.

“One of the major revelations of comparative genomics has been the discovery of regions of the genome falling well outside protein-coding genes that nonetheless exhibit considerable levels of conservation across evolutionary time”

• cites Bejerano et al. 2004 sourceNot checked — no full text of the source
• cites Siepel et al. 2005 sourceSupported
• cites Woolfe et al. 2005 sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Chikina et al. quantified the difference of relative substitution rates between terrestrial and marine branches using a nonparametric test (Wilcoxon rank sum test); the same approach has been extended to noncoding regions.

“Chikina et al. (2016), studying protein-coding genes with convergent substitution rate shifts in marine mammals, quantified the difference of relative substitution rates between “terrestrial” and “marine” branches using a nonparametric test (Wilcoxon rank sum test); the same approach has subsequently been extended to noncoding regions (Partha et al. 2017).”

• cites Chikina et al. 2016 sourceNot checked — no full text of the source
• cites Partha et al. 2017 sourceSupported

Classified as probably not a substantial claim. May combine two claims.

The random local clock model implemented in BEAST assumes an indicator variable for rate change at each node, enumerates all possible local clock configurations, and estimates the location and magnitude of rate changes between local clock regions on the phylogeny.

“The random local clock model implemented in BEAST (Drummond and Suchard 2010) assumes an indicator variable for rate change at each node, enumerating all possible local clock configurations, and estimates the location and magnitude of rate changes between local clock regions on the phylogeny.”

• cites Drummond and Suchard 2010 sourcePartly supported

  Split into 2 atoms — 1 confirmed, 1 partially supported, 0 disputed, 0 misattributed, 0 unverifiable. Expand the claim to see per-atom verdicts.

  Experimental — accuracy not yet measured

Classified as probably not a substantial claim. May combine two claims.

GC-biased gene conversion (gBGC) is one of the factors that can increase local substitution rates across the genome and is often a confounding factor for the detection of adaptive evolution.

“GC-biased gene conversion (gBGC) is one of the factors that can increase local substitution rates across the genome and is often a confounding factor for the detection of adaptive evolution (Duret and Galtier 2009; Kostka et al. 2012; Capra et al. 2013).”

• cites Duret and Galtier 2009 sourceNot checked — no full text of the source
• cites Kostka et al. 2012 sourceSupported
• cites Capra et al. 2013 sourcePartly supported

  The Results (Probabilistic Model; no page given) state that gBGC increases W→S substitution rates, supporting the local-rate claim. The Results (Overlap with Fast-Evolving Sequences; no page given) report overlap between gBGC tracts and some genes identified as showing signatures of positive selection. This supports a potential confounding effect, but does not establish that gBGC often confounds the detection of adaptive evolution. (LLM, conf=0.82)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

Heterogeneity in the topology of gene trees is expected to occur, especially during rapid radiations.

“Heterogeneity in the topology of gene trees is expected to occur, especially during rapid radiations (Edwards 2009).”

• cites Edwards 2009 sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Bayesian multispecies coalescent models like *BEAST and starBEAST2 consider gene tree variations and can give more accurate estimate of per-species substitution rates.

“Bayesian multispecies coalescent models like *BEAST and starBEAST2 consider gene tree variations and can give more accurate estimate of per-species substitution rates (Heled and Drummond 2010; Ogilvie et al. 2017).”

• cites Heled and Drummond 2010 sourcePartly supported

  The Introduction and Comparison with BEST section say that *BEAST jointly estimates multiple gene trees under a multispecies coalescent model, supporting the gene-tree-variation part of the claim. In Results, “Estimation of Relative Substitution Rates for Different Loci” reports estimates of rates across loci, not more accurate estimates of per-species substitution rates. This source describes *BEAST; it does not assess starBEAST2. (LLM, conf=0.90)

  Experimental — accuracy not yet measured
• cites Ogilvie et al. 2017 sourcePartly supported

  The Introduction identifies *BEAST and StarBEAST2 as Bayesian multispecies coalescent methods, and the Conclusions say StarBEAST2’s new species-tree relaxed clock enables accurate inference of per-species branch rates despite incomplete lineage sorting. But the Species Tree Relaxed Clocks section says *BEAST’s gene-tree relaxed clocks do not produce species-branch-rate estimates. The claim therefore overextends the rate-estimation finding to *BEAST. Page numbers are not available in our copy. (LLM, conf=0.87)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

Our model is more specific in its goals than those implemented in more general phylogenetic packages.

“Without doubt, our model is more specific in its goals than those implemented in more general phylogenetic packages (Drummond et al. 2006; Drummond and Suchard 2010; Heath et al. 2012; Ogilvie et al. 2017).”

• cites Drummond et al. 2006 sourceSupported
• cites Drummond and Suchard 2010 sourceSupported
• cites Heath et al. 2012 sourcePartly supported

  The DPP on Branch Rates section identifies a phylogenetic model implemented in the C++ program DPPDiv, so the paper supports the citation as an example of another implemented model. It does not establish that DPPDiv is a general phylogenetic package; that description is not supported for this citation. (LLM, conf=0.84)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

Misspecifying the phylogenetic tree can lead to misestimation of substitution rates on gene trees.

“Misspecifying the phylogenetic tree can lead to misestimation of substitution rates on gene trees (Angelis and dos Reis 2015; Hahn and Nakhleh 2016; Mendes and Hahn 2016).”

• cites Angelis and dos Reis 2015 sourceNot checked — no full text of the source
• cites Hahn and Nakhleh 2016 sourceNot checked — no full text of the source
• cites Mendes and Hahn 2016 sourceSupported

May combine two claims.

Both PhyloAcc and phyloP perform better with longer sequences, and PhyloAcc is better for all lengths in most cases.

“Both methods perform better with longer sequences and PhyloAcc is better for all lengths in most cases.”

May combine two claims.

phyloP performs similarly or better than PAML+Wilcoxon and PAML+phylANOVA.

“Because phyloP performs similarly or better than PAML+Wilcoxon and PAML+phylANOVA, we mainly focus on comparisons with phyloP”

May combine two claims.

In practice, most branches in the accelerated state have estimated values of ri2 > 1, with only a small percentage having values < 1.

“In practice, most branches in the accelerated state do indeed have estimated values of ri2> 1, with only a small percentage having values < 1.”

• AssumptionThe authors state this as an assumption: For any element, most branches are in the conserved state.

Probably not a substantial claim: may not pass S3 (load-bearing).

The average proportion of nucleotide differences per element between pairs of species varies from 0.06 in case 1 to 0.13 in case 8.

“The average proportion of nucleotide differences per element between pairs of species varies from 0.06 (case 1) to 0.13 (case 8).”

The quote only partly states this claim.

PhyloAcc reveals numerous conserved nonexonic elements with accelerations specific to the phenotypically convergent lineages in loss of flight in birds and transition to marine life in mammals.

“In each case, our approach reveals numerous examples of conserved nonexonic elements with accelerations specific to the phenotypically convergent lineages.”

May combine two claims.

In additional simulations, PhyloAcc has a lower false positive rate and higher power in identifying elements with a shift in substitution rate within a set of species.

“In additional simulations (supplementary fig. S5A, Supplementary Material online), we show that our method has a lower false positive rate and a higher power in identifying elements with a shift in substitution rate within a set of species”

May combine two claims.

In both the loss of flight in birds and transition to marine life in mammals, PhyloAcc identified conserved elements with specific patterns of convergent rate shifts associated with the target phenotype, revealing novel putative regulatory regions that may be repeatedly associated with these evolutionary transitions.

“In both cases, we use genome-wide data from hundreds of thousands of conserved elements to identify those elements with specific patterns of convergent rate shifts associated with our target phenotype, revealing novel, putative regulatory regions that may be repeatedly associated with these evolutionary transitions.”

May combine two claims.

In simulations mixing accelerated elements with nonaccelerated elements, PhyloAcc consistently performed best at detecting elements accelerated among ratites based on BF1, except in the ostrich-accelerated case (case 3), where phyloP was best.

“PhyloAcc consistently performed the best (except for the ostrich-accelerated case [case 3], in which phyloP is the best) at detecting elements accelerated among ratites based on BF1”

May combine two claims.

With the second choice of accelerated branches (all subtrees within target lineages inferred by parsimony), phyloP and PhyloAcc have comparable performances, and phyloP is only better for shorter sequences in some cases.

“With the second choice of accelerated branches, phyloP and PhyloAcc have comparable performances and phyloP is only better for shorter sequence in some cases”

May combine two claims.

BEAST2 has comparable accuracy when no clade or a large clade is accelerated, but performs worse than PhyloAcc in cases with multiple independent rate shifts (e.g., cases 7 and 9) or with rate shifts on short branches (e.g., case 2).

“BEAST2 has a comparable accuracy when no clade or a large clade is accelerated, but performs worse than PhyloAcc in cases with multiple independent rate shifts (e.g., cases 7 and 9) or with rate shifts on short branches (e.g., case 2).”

May combine two claims.

Marine-accelerated CNEEs show acceleration in only a small number of nontargeted species, and fewer than control-accelerated CNEEs.

“We observed acceleration in only a small number of nontargeted species for marine-accelerated CNEEs and fewer than that in control-accelerated CNEEs”

May combine two claims.

PhyloAcc can identify both elements 1 and 2 as ratite-accelerated elements, indicated by large log-BF1, but can exclude element 3 because it has a negative log-BF2.

“In contrast, PhyloAcc can identify both elements 1 and 2 as ratite-accelerated elements, indicted by large log-BF1, but can also exclude element 3 because it has a negative log-BF2.”

May combine two claims.

For PhyloAcc, BF1 increases with the rate ratio, and BF2 stays below zero and decreases as r2/r1 grows in cases 8 and 9.

“In accordance with accuracy, for PhyloAcc, BF1 also increases with the rate ratio, and BF2 stays below zero and decreases as r2/r1 grows in cases 8 and 9”

May combine two claims.

Among the 786 ratite-specific accelerated CNEEs, the rhea clade is the most likely lineage to be accelerated, followed by kiwis, with the ostrich branch less likely accelerated among all ratites.

“The rhea clade is the most likely lineage to be accelerated among the 786 ratite-specific accelerated CNEEs, followed by kiwis, with the ostrich branch less likely accelerated among all ratites”

May combine two claims.

Among CNEEs with strong signals of acceleration in ratites indicated by very large Bayes factors (BF1), many are accelerated in a single clade with one acceleration, whereas some are accelerated in more than one clade.

“Among CNEEs that show strong signals of acceleration in ratites indicated by very large Bayes factors (BF1), many are accelerated in a single clade with one acceleration (e.g., mCE600387 accelerated only in kiwis, fig. 4B) whereas some are accelerated in more than one clade”

May combine two claims.

93 (4.4%) of marine-accelerated elements show more than two independent accelerations compared with 33 (2.2%) for control-accelerated elements.

“93 (4.4%) of marine-accelerated elements show more than two independent accelerations compared with 33 (2.2%) for control-accelerated elements”

Classified as probably not a substantial claim.

Molecular clock tests are closely linked to tests for associations between rate shifts and specific convergent phenotypes on the tree.

“Molecular clock tests are closely linked to tests for associations between rate shifts and specific convergent phenotypes on the tree, and several recent methods have been proposed to identify these associations.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

Existing methods do not allow easy detection of patterns of rate variation, which causes challenges for detecting convergent rate shifts or other complex evolutionary scenarios.

“However, existing methods do not allow easy detection of patterns of rate variation, which causes challenges for detecting convergent rate shifts or other complex evolutionary scenarios.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Such pooling plays a larger role in cases where only a few branches are accelerated, and/or few substitutions occur per element.

“Such pooling plays a larger role in cases where only a few branches are accelerated, and/or few substitutions occur per element.”

• AssumptionThe authors state this as an assumption: A Gamma distribution is an appropriate prior for element-wise substitution rates (ri1 and ri2).
• AssumptionThe authors state this as an assumption: Most elements in the data set likely do not change rate across the tree.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Statistical tests for rate changes along a phylogeny have been a part of phylogenetic methodology for several decades and are closely tied to tests for a molecular clock.

“Statistical tests for rate changes along a phylogeny have been a part of phylogenetic methodology for several decades, and are closely tied to tests for a molecular clock”

• cites Huelsenbeck and Crandall 1997 sourceNot checked — no full text of the source
• cites Huelsenbeck and Rannala 1997 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

The model allows a variety of scenarios whose biological interpretation might include partially relaxed constraint or positive selection with a rate greater than background.

“This allows us to model a variety of scenarios whose biological interpretation might include partially relaxed constraint or positive selection with a rate greater than background.”

• AssumptionThe authors state this as an assumption: For each element, a limited number of rate categories occur on the phylogeny.

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

PhyloAcc is widely applicable to any set of conserved elements where multiple rate changes are expected on a phylogeny.

“Our method is widely applicable to any set of conserved elements where multiple rate changes are expected on a phylogeny.”

Classified as probably not a substantial claim.

Bayes factors favor the simpler model if both models fit the data equally well.

“because the Bayes factors favor the simpler model if both models fit the data equally well”

May combine two claims.

Existing methods for detecting genomic regions with parallel substitution rate changes are generally limited to testing a single prespecified shift pattern, which is less powerful because many possible shift patterns would be congruent with a correlation with phenotypes of extant species.

“However, these methods for detecting genomic regions with parallel substitution rate changes are generally limited to testing a single prespecified shift pattern on a phylogeny, which is less powerful because many possible shift patterns would be congruent with a correlation with phenotypes of extant species.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Noncoding elements play a crucial role in molding morphological diversity across diverse clades.

“Such studies demonstrate that noncoding elements play a crucial role in molding morphological diversity across diverse clades.”

Classified as probably not a substantial claim. May combine two claims.

The posterior distribution of Z indicates where changes of substitution rates occur in the tree, and the posterior ratio of substitution rates for each latent state indicates the magnitude of change.

“The posterior distribution of Z indicates where changes of substitution rates occur in the tree, and the posterior ratio of substitution rates for each latent state indicates the magnitude of change.”

• AssumptionThe authors state this as an assumption: The substitution process follows a standard continuous-time Markov process.
• AssumptionThe authors state this as an assumption: For each element, a limited number of rate categories occur on the phylogeny.

May combine two claims.

PhyloAcc does not require predetermination of the history of rate shifts but instead relies on estimating the conservation state of each branch for a given element based on sequences of extant species.

“PhyloAcc does not require predetermination of the history of rate shifts but instead relies on estimating the conservation state of each branch for a given element based on sequences of extant species.”

• AssumptionThe authors state this as an assumption: The substitution process follows a standard continuous-time Markov process.
• AssumptionThe authors state this as an assumption: The substitution rate of a given branch is expected to correlate with that of its parent branch.
• AssumptionThe authors state this as an assumption: For each element, a limited number of rate categories occur on the phylogeny.

May combine two claims.

Other methods are not designed to control for acceleration in nontarget lineages, and their test statistics can be even larger in case 8 than in some target-specific accelerated cases.

“Other methods are not designed to control for this case; indeed the test statistic from phyloP or the two-step methods can be even larger in case 8 than in some target-specific accelerated cases.”

May combine two claims.

Changes in evolutionary rates on particular lineages in a phylogeny can indicate shared functional shifts and can be used to detect genomic correlates of phenotypic convergence.

“Changes in evolutionary rates on particular lineages in a phylogeny can indicate shared functional shifts, and thus can be used to detect genomic correlates of phenotypic convergence.”

Classified as probably not a substantial claim.

By adjusting the hyperparameters of the priors, the substitution rate for the accelerated state can be made stochastically higher than that of the conserved state a priori.

“By adjusting the hyperparameters of the priors, the substitution rate for the accelerated state can be made stochastically higher than that of conserved state a priori.”

• AssumptionThe authors state this as an assumption: A Gamma distribution is an appropriate prior for element-wise substitution rates (ri1 and ri2).

May combine two claims.

The resulting output can be used to reconstruct the sequence of shifts in rate, including the number of independent accelerations of a particular element, and quantifies the uncertainty of where in the tree accelerations occur.

“The resulting output can be used to reconstruct the sequence of shifts in rate, including the number of independent accelerations of a particular element, and quantifies the uncertainty of where in the tree (on which branch) accelerations occur.”

• AssumptionThe authors state this as an assumption: The substitution process follows a standard continuous-time Markov process.
• AssumptionThe authors state this as an assumption: For each element, a limited number of rate categories occur on the phylogeny.

Probably not a substantial claim: may not pass S3 (load-bearing).

The PHAST method tests a model allowing substitution rates shift in a specified subset of branches against a null model with constant rate for all branches using likelihood ratio.

“The PHAST method (Hubisz et al. 2011) tests a model allowing substitution rates shift in a specified subset of branches against a null model with constant rate for all branches using likelihood ratio.”

• cites Hubisz et al. 2011 sourceNot checked — no full text of the source

May combine two claims.

PhyloAcc outputs the most probable evolutionary pattern of rate shifts as well as its uncertainty for each genomic region.

“Using Markov chain Monte Carlo (MCMC) (Liu 2001) to sample from the posterior distribution, PhyloAcc outputs the most probable evolutionary pattern of rate shifts as well as its uncertainty for each genomic region.”

• cites Liu 2001Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• AssumptionThe authors state this as an assumption: The substitution process follows a standard continuous-time Markov process.
• AssumptionThe authors state this as an assumption: For each element, a limited number of rate categories occur on the phylogeny.

May combine two claims.

Via Dollo's assumption of irreversibility, PhyloAcc allows at most two shifts for each lineage on the phylogeny, which may not be efficient for detecting elements that regain conservation after an ancient episode of adaptation.

“In addition, via Dollo’s assumption of irreversibility, our model allows at most two shifts for each lineage on the phylogeny, which may not be efficient for detecting elements that regain conservation after an ancient episode of adaptation.”

• AssumptionThe authors state this as an assumption: Conservation states evolve under Dollo’s assumption of irreversibility, so a lineage can shift at most twice.

May combine two claims.

As a consequence, BEAST2 tends to misplace the origin of an acceleration at a node deeper than the true node and then infer a regain of conservation in the clade whose rate is unchanged.

“As a consequence, BEAST2 tends to misplace the origin of an acceleration at a node deeper than the true node and then infer a regain of conservation in the clade whose rate is unchanged.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The conserved rate r1 and accelerated rate r2 used in simulations were sampled from Gamma distributions that approximate the range of conserved and accelerated rates observed in real data.

“The conserved rate r1 was sampled from Gamma(5,0.04) and the accelerated rate r2 was sampled from Gamma(15,0.1), which are about the range of conserved and accelerated rates observed from real data (supplementary figs. S5C and S12D, Supplementary Material online).”

May combine two claims.

The core utility of PhyloAcc is to detect changes of substitution rate on a tree for many conserved DNA elements, yielding the posterior distribution of substitution rates per conservation state (r1 and r2) for each element on each branch, from which the direction of rate change can be inferred.

“The core utility of our software PhyloAcc is to detect changes of substitution rate on a tree for many conserved DNA elements, yielding the posterior distribution of substitution rates per conservation state (i.e., r1 and r2) for each element on each branch, from which the direction of rate change can be inferred.”

May combine two claims.

Because scores depend on both magnitude and pattern of acceleration, elements with the same log-likelihood ratios and P-values can have either strong acceleration in a single lineage, multiple independent accelerations, or weak acceleration over an entire clade.

“As these scores depend on both the magnitude and pattern of acceleration, it is not hard to find elements with the same log-likelihood ratios and P-values having either strong acceleration in a single lineage, multiple independent accelerations or a weak acceleration over an entire clade.”

Borderline on S4 (based): unclear from the text.

The likelihood ratio test between constant rate and two-rate models is not appropriate in some extreme cases, such as when only a few branches have many substitutions.

“The likelihood ratio test between constant rate and two-rate models is not appropriate in some extreme cases, such as when only a few branches have many substitutions.”

Probably not a substantial claim: may not pass S3 (load-bearing).

As currently implemented, the method is efficiently tailored to the one goal of detecting rate shifts in genome-wide noncoding elements on a phylogenetic tree.

“As currently implemented, it is efficiently tailored to the one goal of detecting rate shifts in genome-wide noncoding elements on a phylogenetic tree.”

May combine two claims.

The conservation state of short internal branches is easier to determine when more substitutions are observed, which tends to occur when accelerated rates are high.

“because the conservation state of short internal branches is easier to determine when we observe more substitutions, which will tend to occur when accelerated rates are high.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Another major issue not addressed by our model is the possible heterogeneity in the topologies of gene trees across elements and across the genome.

“Another major issue not addressed by our model is the possible heterogeneity in the topologies of gene trees across elements and across the genome.”

• AssumptionRests on an assumption: The phylogenetic tree topology is the same for all elements and across the genome (no gene tree heterogeneity).

Probably not a substantial claim: may not pass S3 (load-bearing).

CLIME (Li et al. 2014, 2018) uses a Bayesian approach similar in spirit to our model to cluster phylogenetic profiles of proteins to identify functionally related proteins.

“For example, a recent method, “CLIME” (Li et al. 2014, 2018), uses a Bayesian approach similar in spirit to our model to cluster phylogenetic profiles of proteins to identify functionally related proteins.”

• cites Li et al. 2014, 2018Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

Many of the enriched functional terms are related to morphological traits, which reveals molecular adaptations overlooked by previous studies that focused primarily on protein-coding genes.

“Many of the enriched functional terms are related to morphological traits, which reveals molecular adaptations overlooked by previous studies, which focused primarily on protein-coding genes.”

May combine two claims.

PhyloAcc identifies DNA elements exhibiting changes of substitution rate in lineages of interest and determines the branches, leading to either single or multiple lineages, experiencing changes of substitution rate.

“The method not only identifies DNA elements exhibiting changes of substitution rate in the lineages of interest but also determines the branches, leading to either single or multiple lineages, experiencing changes of substitution rate”

Classified as probably not a substantial claim.

We do not attempt to distinguish among various processes-different types of mutation, increased mutation rates or natural selection-producing a specific pattern.

“Although in our examples we focus on loss of conservation accompanied by faster substitution rates, we do not attempt to distinguish among various processes—different types of mutation, increased mutation rates or natural selection—producing a specific pattern.”