Eukaryotic protein-coding genes are interrupted by introns, which have to be excised from the primary transcript to produce functional mRNAs that can be translated into proteins. The removal of introns from primary transcripts can lead to the production of diverse mRNAs, via the differential use of splice sites. This process of alternative splicing (AS) is widespread in eukaryotes (Chen et al., 2014), but its ’raison d’être’ (adaptive or not) remains elusive. Numerous studies have shown that some AS events are functional, that is that they play a beneficial role for the fitness of organisms, either by allowing the production of distinct protein isoforms (Graveley, 2001) or by regulating gene expression post-transcriptionally (McGlincy and Smith, 2008; Hamid and Makeyev, 2014). However, other AS events are undoubtedly not functional. Like any biological machinery, the spliceosome occasionally makes errors, leading to the production of aberrant mRNAs, which represent a waste of resources and are therefore deleterious for the fitness of the organisms (Hsu and Hertel, 2009; Gout et al., 2013). The splicing error rate at a given intron is expected to depend both on the efficiency of the spliceosome and on the intrinsic quality of its splice signals. The information required in cis for the removal of each intron resides in 20–40 nucleotide sites, located within the intron or its flanking exons (Lynch, 2006). Besides the two splice sites that are essential for the splicing reaction (almost always GT for the donor and AG for the acceptor), all other signals tolerate some sequence flexibility. Population genetics principles state that the ability of selection to promote beneficial mutations or eliminate deleterious mutations depends on the intensity of selection (s) relative to the power of random genetic drift (defined by the effective population size, Ne): if the selection coefficient is sufficiently weak relative to drift (${\displaystyle |N_{\text{e}}s|\ll 1}$), alleles behave as if they are effectively neutral. Thus, random drift sets an upper limit on the capacity of selection to prevent the fixation of alleles that are sub-optimal (Kimura et al., 1963; Ohta, 1973). This so-called ‘drift barrier’ (Lynch, 2007) is expected to affect the efficiency of all cellular processes, including splicing. Hence, species with low Ne should be more prone to make splicing errors than species with high Ne.

The extent to which AS events correspond to functional isoforms or to errors is a contentious issue (Bhuiyan et al., 2018; Tress et al., 2017b; Blencowe, 2017; Tress et al., 2017a). In humans, the set of transcripts produced by a given gene generally consists of one major transcript (the ‘major isoform‘), which encodes a functional protein, and of multiple minor isoforms (splice variants), present in relatively low abundance, and whose coding sequence is frequently interrupted by premature termination codons (PTCs) (Tress et al., 2017a; Gonzàlez-Porta et al., 2013). Ultimately, less than 1% of human splice variants lead to the production of a detectable amount of protein (Abascal et al., 2015). Furthermore, comparison with closely related species showed that AS patterns evolve very rapidly (Barbosa-Morais et al., 2012; Merkin et al., 2012) and that alternative splice sites present little evidence of selective constraints (Pickrell et al., 2010). All these observations are consistent with the hypothesis that a vast majority of splice variants observed in human transcriptomes simply correspond to erroneous transcripts (Pickrell et al., 2010). However, some authors argue that a large fraction of AS events might in fact contribute to regulating gene expression. Indeed, PTC-containing splice variants are recognized and degraded by the non-sense mediated decay (NMD) machinery. Thus, AS can be coupled with NMD to modulate gene expression at the post-transcriptional level (McGlincy and Smith, 2008; Hamid and Makeyev, 2014). This AS-NMD regulatory process does not involve the production of proteins and does not necessarily imply strong evolutionary constraints on splice sites. Thus, based on these observations, it is difficult to firmly refute selectionist or non-adaptive models.

The analysis of transcriptomes from various eukaryotic species showed substantial variation in AS rates across lineages, with the highest rate in primates (Barbosa-Morais et al., 2012; Chen et al., 2014; Mazin et al., 2021). Interestingly, the genome-wide average AS level was found to correlate positively with the complexity of organisms (approximated by the number of cell types; Chen et al., 2014). This correlation was considered as evidence that AS contributed to the evolution of complex organisms by increasing the functional repertoire of their genomes (Chen et al., 2014). This pattern is often presented as an argument supporting the importance of AS in adaptation (Verta and Jacobs, 2022; Singh and Ahi, 2022; Wright et al., 2022). However, this correlation is also compatible with a totally opposite hypothesis. Indeed, eukaryotic species with the highest level of complexity correspond to multi-cellular organisms with relatively large body size, which tend to have small effective population sizes (Ne) (Lynch and Conery, 2003; Figuet et al., 2016). Thus, the higher AS rate observed in ‘complex’ organisms might simply reflect an increased rate of splicing errors, resulting from the effect of the drift barrier on the quality of splice signals (Bush et al., 2017).

To assess this hypothesis and evaluate the impact of genetic drift on alternative splicing patterns, we quantified AS rates in 53 metazoan species, covering a wide range of Ne values, and for which high-depth transcriptome sequencing data were available. We show that the genome-wide average AS rate correlates negatively with Ne, in agreement with the drift barrier hypothesis. This pattern is mainly driven by low abundance isoforms, which represent the vast majority of splice variants and most likely correspond to errors. Conversely, the AS rate of abundant splice variants, which are enriched in functional AS events, show the opposite trend. These results support the hypothesis that the drift barrier sets an upper limit on the capacity of selection to minimize splicing errors.

To investigate the factors that drive variation in AS rates across species, we analyzed publicly available RNA-seq data across a large set of 53 species, from diverse metazoan clades, covering a wide range of Ne values. To facilitate comparisons across species, we sought to limit the impact of the among-gene variance in AS rates. For this, we primarily based our analyses on a common set of nearly 1000 orthologous protein-coding genes (BUSCO gene set). We focused our study on introns located within protein-coding regions, because introns from UTRs or lncRNAs are expected to be subject to different functional constraints. We measured AS rates on introns corresponding to a major isoform. When sequencing depth is limited, the set of introns for which AS can be quantified is biased toward the most highly expressed genes. To avoid this bias, we restricted our study to species for which the median sequencing depth of BUSCO exons was above 200. With this setting, on average 96.9% of BUSCO annotated introns could be analyzed in each species (Figure 1—source data 1).

We observed a fivefold variation in the average AS rate of BUSCO introns across species from 0.8% in Drosophila grimshawi (Diptera) to 3.8% in Megachile rotundata (Hymenoptera)(Figure 3A). In agreement with previous work, we observed that AS rates tend to be high in vertebrates (average = 2.3%), and notably in primates (average = 3.1%) (Barbosa-Morais et al., 2012; Chen et al., 2014; Mazin et al., 2021). This observation was previously interpreted as an evidence that AS played an important role in the diversification of the functional repertoire necessary for the development of more complex organisms (Chen et al., 2014). However, this pattern is also compatible with the hypothesis that variation in AS rates across species result from differences in splicing error rates, which are expected to be higher in species with low Ne (Bush et al., 2017). Indeed, consistent with this drift barrier hypothesis, we observed significant correlations between AS rates and proxies of Ne (Figure 3B, Figure 3—figure supplement 1A and B).

In their original study, Chen et al., 2014 investigated the hypothesis that variation in AS rates across taxa might be driven by variation in Ne. For this, they focused on 12 species, for which they had measured levels of polymorphism at silent sites (π). They found that the correlation between AS rate and the number of cell types (proxy for organismal complexity) remained significant after controlling for π. They therefore concluded that the association between the cellular diversity and alternative splicing was not a by-product of reduced effective population sizes among more complex species. This conclusion was however based on a very small sample of species. More importantly, it assumed that π could be taken as a proxy for Ne. At mutation-drift equilibrium, π is expected to be proportional to Neu (where u is the mutation rate per bp per generation). Thus, if u is constant across taxa, π can be used to estimate variation in Ne. However, the dataset analyzed by Chen et al., 2014 included very diverse eukaryotic species, with mutation rates ranging from 1.7 x10¹⁰ mutation per bp per generation in budding yeast, to 1.1 x10⁸ mutation per bp per generation in humans (Lynch et al., 2016). Hence, at this evolutionary scale, variation in Ne cannot be directly inferred from π without accounting for variation in u. Moreover, the drift barrier hypothesis states that the AS rate of a species should reflect the genome-wide burden of slightly deleterious substitutions, which is expected to depend on the intensity of drift over long evolutionary times (i.e. long-term Ne). Conversely, π reflects Ne over a short period of time (of the order of Ne generations), and can be strongly affected by recent population bottlenecks (too recent to have substantially impacted the genome-wide deleterious substitution load). The drift barrier hypothesis therefore predicts that the splicing error rate should correlate more strongly with proxies of long-term Ne (such as ${\displaystyle dN/dS}$, life history traits, or organismal complexity) than with π. The fact that AS rates remained significantly correlated to cellular diversity after controlling for π(Chen et al., 2014) is therefore not a conclusive argument against the drift barrier hypothesis.

To contrast the two models (drift barrier vs diversification of the functional repertoire in complex organisms), we sought to distinguish functional splice isoforms from erroneous splicing events. Based on the assumption that splicing errors should occur at a low frequency, we split major-isoform introns into two categories, those with abundant SVs (${\displaystyle \mathrm{M}\mathrm{I}\mathrm{R}\mathrm{A}>5}$%), and those without (${\displaystyle \mathrm{M}\mathrm{I}\mathrm{R}\mathrm{A}\le 5}$%). Rare SVs represent the vast majority of the repertoire of splicing isoforms detected in a given transcriptome (from 62.4% to 96.9% according to the species; Figure 1—source data 1). Two lines of evidence indicate that the small subset of abundant isoforms is strongly enriched in functional transcripts relative to other SVs. First, we observed that in all species, the proportion of SVs that preserve the reading frame is much higher among abundant SVs than among rare SVs (Figure 4A). Second, the analysis of polymorphism data in Drosophila indicates that the average level of purifying selection on alternative splice sites is much stronger for abundant than rare SVs (Figure 5A and B).

If variation in AS rate across species had been driven by a higher prevalence of functional SVs in more complex organisms, one would have expected the proportion of frame-preserving SVs to be stronger in vertebrates than in insects, in particular for the set of introns with high AS rate (i.e. enriched in functional SVs). On the contrary, the highest proportion of frame-preserving SVs is observed in dipterans (Figure 4B). In fact, the overall higher AS rate of vertebrates (Figure 3A) is driven by the set of introns with a low AS rate (Figure 4C), that is the set of introns in which the prevalence of functional SVs is the lowest. On the contrary, among the set of introns with high AS rate, vertebrates have lower AS rates than insects (Figure 4D).

These observations are difficult to reconcile with the hypothesis that the higher AS rate in vertebrates results from a higher rate of functional AS. Conversely, these observations fit very well with a model where variation in AS rate across species is entirely driven by variation in the efficacy of selection against splicing errors. To illustrate this model, let us consider three hypothetical species with different Ne, in which a small fraction of major-isoform introns (say 5%) is subject to functional alternative splicing. Let us consider that the distribution of AS rates of functional splicing variants is the same for all species (i.e. independent of Ne), with a mean of 25% (and a standard deviation of 5%). In addition, we assume that all major-isoform introns are potentially affected by splicing errors, with a mean error rate ranging from 0.2% in species of high Ne to 1.2% in species of low Ne, owing to the drift barrier effect (these parameters were set to match approximately the AS rates observed in empirical data for rare SVs). The distributions of AS rate given by this model are presented in Figure 7A: rare SVs (MIRA ≤ 5%) essentially correspond to splicing errors, while abundant SVs (MIRA > 5%) correspond to a mixture of functional and spurious variants, whose relative proportion depend on Ne (Figure 7B). This simple model makes predictions that match with our observations: we noted a positive correlation between AS rate and longevity (i.e. a negative correlation with Ne) for the set of low-AS major-isoform introns (Figure 4C), but an opposite trend for high-AS major-isoform introns (Figure 4D), as predicted by the model (Figure 7D and E). Given that high-AS major-isoform introns represent only a small fraction of major-isoform introns, this model predicts that, overall, AS rates correlate negatively with Ne (Figure 7), as observed in empirical data (Figure 3A, Figure 3—figure supplement 1).

Figure 7

Download asset Open asset

It should be noted that the BUSCO dataset corresponds to genes that are strongly conserved across species, often highly expressed, and hence might not be representative of the entire genome. Notably, AS rates are on average lower in the BUSCO gene set than in other genes, even after accounting for their expression level (Figure 6—figure supplement 1). However, results remained qualitatively unchanged when we repeated our analyses on the whole set of annotated protein-coding genes for each species: correlations between AS rates and Ne proxies are slightly weaker than on the BUSCO subset, but remain significant (Figure 4—figure supplement 1).

The model also predicts that the proportion of functional SVs among high-AS major-isoform introns should vary with Ne (Figure 7C). To assess this point, we measured in each species the enrichment in reading frame-preserving events among abundant SVs compared to rare SVs. As predicted, this estimate of the prevalence of functional SVs tends to decrease with decreasing Ne proxies (e.g. Figure 4B, where Ne is approximated by longevity). However, these correlations are weak, marginally significant after accounting for phylogenetic inertia with only two of the three Ne proxies, and not robust to multiple testing issues (Figure 4—figure supplement 2). Thus, Ne does not appear to be a strong predictor of the prevalence of functional SVs among high-AS major-isoform introns.

According to the drift-barrier model, the level of splicing errors is expected to decrease with increasing selective pressure. In all above analyses, we considered AS rates measured per intron, and not per gene. Yet, the trait under selection is the per-gene error rate, which depends not only on the error rate per intron, but also on the number of introns per gene. Given that intron density varies widely across clades (from 2.8 introns per gene in diptera to 8.4 introns per gene in vertebrates; ), the correlations reported above between AS rates and Ne may undervalue the predictive power of the drift-barrier model. The RNA-seq datasets that we analyzed consist of short-read sequences, which do not allow a direct quantification of the per-gene AS rate. We therefore indirectly estimated the per-gene AS rate in each species, based on the per-intron AS rate and on the number of introns per gene (Materials and methods). Interestingly, as predicted by the drift-barrier model, Ne proxies correlate more strongly with this estimate of the per-gene AS than with the per-intron AS rates (Figure 3—figure supplement 4).

One other important prediction of the drift barrier model is that splicing error rate should vary not only across species according to Ne, but also among genes, according to their expression level. Indeed, for a given splicing error rate, the waste of resources (and hence the fitness cost) is expected to increase with the level of transcription. Thus, the selective pressure for optimal splice signals is expected to be higher, and hence the error rate to be lower, in highly expressed genes. Consistent with that prediction, we observed a negative correlation between gene expression level and AS rate in low-AS major-isoform introns in all but one species (Figure 6C).

It should be noted that our analyses suffer from several important limitations. First, the proxies that we considered for Ne are quite noisy (Figure 1). Second, to maximize the number of species in our analyses, we had to use very heterogeneous sources of RNA (whole-body, specific tissues, or organs, at different life stages, in different sexes, different environmental conditions, etc.). Third, we used short-read sequencing data, which allow the quantification of AS rates for individual introns, but do not provide a direct measure of AS rates per gene. Hopefully progress of long-read sequencing technologies will soon allow the comparative analysis of AS rates on full-length transcripts (e.g. see Leung et al., 2021). But presently, publicly available long-read transcriptomic data are restricted to a narrow set of model organisms, and their sequencing depth is still too limited to quantify rare splicing events. The fact that we detected significant correlations between AS rate and the three Ne proxies, despite these uncontrolled sources of variability, suggests that we underestimate the effect of Ne on AS rates.

Thus, overall, all observations fit qualitatively well with the predictions of the drift barrier model, according to which most of the variation in AS rate across species reflects differences in splicing error rates. Of course, this model is not in contradiction with the fact, well established, that some AS events play an essential role in various processes. Different criteria can be used to distinguish functional SVs from spurious splicing events. Notably, AS events that are strongly tissue-specific or developmentally dynamic tend to be more conserved across species, which indicates that a substantial fraction of them are evolutionary constrained, and hence functional (Mudge et al., 2011; Barbosa-Morais et al., 2012; Merkin et al., 2012; Reyes et al., 2013). The abundance of an SV is also an important predictor of its functionality. In particular, we observed that in all species, the proportion of frame-preserving events is much higher among abundant SVs than among rare SVs (Figure 4A). We note however that the threshold that we used to define abundant SVs is somewhat arbitrary. In fact, according to our model, this class of SVs corresponds to a mixture of functional and spurious events, whose relative proportion is expected to depend on Ne (Figure 7C). Thus, in low-Ne species, even the subset of abundant SVs includes a substantial fraction of errors. This probably explains why, contrarily to Drosophila, we do not detect any signature of purifying selection on alternative splice signals in humans, even for abundant SVs (Figure 5).

In conclusion, all observations fit with the hypothesis that random genetic drift sets an upper limit on the capacity of selection to prevent splicing errors. It should be noted that this limit on the optimization of genetic systems is expected to affect not only splicing, but all aspects of gene expression. Notably, there is a growing body of evidence that the complexity of transcripts produced by eukaryotic genes (resulting from alternative transcription initiation, polyadenylation, splicing or back-splicing, RNA editing) often does not correspond to fine-tuned adaptations but simply to the accumulation of errors (Pickrell et al., 2010; Saudemont et al., 2017; Xu et al., 2019; Xu and Zhang, 2018; Liu and Zhang, 2018b; Liu and Zhang, 2018a; Xu and Zhang, 2014; Xu and Zhang, 2020; Gout et al., 2013; Zhang and Xu, 2022). It should be noted however that the relationship between the genome-wide error rate and Ne is not expected to be monotonic. Indeed, models predict that in species with very high Ne, selection on each individual gene should favor genotypes that are robust to errors of the gene expression machinery, which in turn, reduces the constraints on the global level of gene expression errors (Rajon and Masel, 2011; Xiong et al., 2017). Thus, paradoxically, species with very large Ne are expected to have gene expression machineries that are more error-prone than species with very small Ne (Rajon and Masel, 2011). This argument was developed by Xiong et al., 2017 to account for the fact that transcription error rates had been found to be about 10 times higher in bacteria than in eukaryotes (Traverse and Ochman, 2016; Gout et al., 2013). More recent work indicates that bacterial transcription error rates had been largely overestimated, presumably owing to RNA damages during the preparation of sequencing libraries (Li and Lynch, 2020). Given these uncertainties in the measures of transcription error rates, it seems for now difficult to interpret the differences reported across species. But in any case, it is important to note that it is in principle possible that the drift barrier affects differently the different steps of the gene expression process. It would therefore be important to investigate to which extent each step of gene expression responds (or not) to variation in Ne. As illustrated here by the relationship observed between alternative splicing and Ne, it appears essential to consider the contribution of non-adaptive evolutionary processes when trying to understand the origin of eukaryotic gene expression complexity.

All processed data that we generated and used in this study, as well as the scripts that we used to analyze the data and to generate the figures, are available on zenodo DOI: https://doi.org/10.5281/zenodo.7415114.

1. 1. Bénitière F
   2. Necsulea A
   3. Duret L

   (2024) Zenodo

   2023-Random genetic drift sets an upper limit on mRNA splicing accuracy in metazoans.

   https://doi.org/10.5281/zenodo.7415114

1. 1. Abascal F
   2. Ezkurdia I
   3. Rodriguez-Rivas J
   4. Rodriguez JM
   5. del Pozo A
   6. Vázquez J
   7. Valencia A
   8. Tress ML

   (2015) Alternatively spliced homologous exons have ancient origins and are highly expressed at the protein level

   PLOS Computational Biology 11:e1004325.

   https://doi.org/10.1371/journal.pcbi.1004325

   • PubMed
   • Google Scholar
2. 1. Agarwala R
   2. Barrett T
   3. Beck J
   4. Benson DA
   5. Bollin C
   6. Bolton E
   7. Bourexis D
   8. Brister JR
   9. Bryant SH
   10. Canese K
   11. Cavanaugh M
   12. Charowhas C
   13. Clark K
   14. Dondoshansky I
   15. Feolo M
   16. Fitzpatrick L
   17. Funk K
   18. Geer LY
   19. Gorelenkov V
   20. Graeff A
   21. Hlavina W
   22. Holmes B
   23. Johnson M
   24. Kattman B
   25. Khotomlianski V
   26. Kimchi A
   27. Kimelman M
   28. Kimura M
   29. Kitts P
   30. Klimke W
   31. Kotliarov A
   32. Krasnov S
   33. Kuznetsov A
   34. Landrum MJ
   35. Landsman D
   36. Lathrop S
   37. Lee JM
   38. Leubsdorf C
   39. Lu Z
   40. Madden TL
   41. Marchler-Bauer A
   42. Malheiro A
   43. Meric P
   44. Karsch-Mizrachi I
   45. Mnev A
   46. Murphy T
   47. Orris R
   48. Ostell J
   49. O’Sullivan C
   50. Palanigobu V
   51. Panchenko AR
   52. Phan L
   53. Pierov B
   54. Pruitt KD
   55. Rodarmer K
   56. Sayers EW
   57. Schneider V
   58. Schoch CL
   59. Schuler GD
   60. Sherry ST
   61. Siyan K
   62. Soboleva A
   63. Soussov V
   64. Starchenko G
   65. Tatusova TA
   66. Thibaud-Nissen F
   67. Todorov K
   68. Trawick BW
   69. Vakatov D
   70. Ward M
   71. Yaschenko E
   72. Zasypkin A
   73. Zbicz K
   74. NCBI Resource Coordinators

   (2018) Database resources of the national center for biotechnology information

   Nucleic Acids Research 46:D8–D13.

   https://doi.org/10.1093/nar/gkx1095

   • Google Scholar
3. 1. Auton A
   2. Brooks LD
   3. Durbin RM
   4. Garrison EP
   5. Kang HM
   6. Korbel JO
   7. Marchini JL
   8. McCarthy S
   9. McVean GA
   10. Abecasis GR
   11. 1000 Genomes Project Consortium

   (2015) A global reference for human genetic variation

   Nature 526:68–74.

   https://doi.org/10.1038/nature15393

   • PubMed
   • Google Scholar
4. 1. Barbosa-Morais NL
   2. Irimia M
   3. Pan Q
   4. Xiong HY
   5. Gueroussov S
   6. Lee LJ
   7. Slobodeniuc V
   8. Kutter C
   9. Watt S
   10. Colak R
   11. Kim T
   12. Misquitta-Ali CM
   13. Wilson MD
   14. Kim PM
   15. Odom DT
   16. Frey BJ
   17. Blencowe BJ

   (2012) The evolutionary landscape of alternative splicing in vertebrate species

   Science 338:1587–1593.

   https://doi.org/10.1126/science.1230612

   • PubMed
   • Google Scholar
5. 1. Bhuiyan SA
   2. Ly S
   3. Phan M
   4. Huntington B
   5. Hogan E
   6. Liu CC
   7. Liu J
   8. Pavlidis P

   (2018) Systematic evaluation of isoform function in literature reports of alternative splicing

   BMC Genomics 19:637.

   https://doi.org/10.1186/s12864-018-5013-2

   • PubMed
   • Google Scholar
6. 1. Blencowe BJ

   (2017) The relationship between alternative splicing and proteomic complexity

   Trends in Biochemical Sciences 42:407–408.

   https://doi.org/10.1016/j.tibs.2017.04.001

   • PubMed
   • Google Scholar
7. 1. Bolívar P
   2. Guéguen L
   3. Duret L
   4. Ellegren H
   5. Mugal CF

   (2019) GC-biased gene conversion conceals the prediction of the nearly neutral theory in avian genomes

   Genome Biology 20:5.

   https://doi.org/10.1186/s13059-018-1613-z

   • PubMed
   • Google Scholar
8. 1. Bush SJ
   2. Chen L
   3. Tovar-Corona JM
   4. Urrutia AO

   (2017) Alternative splicing and the evolution of phenotypic novelty

   Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences 372:20150474.

   https://doi.org/10.1098/rstb.2015.0474

   • PubMed
   • Google Scholar
9. 1. Cardoso-Moreira M
   2. Halbert J
   3. Valloton D
   4. Velten B
   5. Chen C
   6. Shao Y
   7. Liechti A
   8. Ascenção K
   9. Rummel C
   10. Ovchinnikova S
   11. Mazin PV
   12. Xenarios I
   13. Harshman K
   14. Mort M
   15. Cooper DN
   16. Sandi C
   17. Soares MJ
   18. Ferreira PG
   19. Afonso S
   20. Carneiro M
   21. Turner JMA
   22. VandeBerg JL
   23. Fallahshahroudi A
   24. Jensen P
   25. Behr R
   26. Lisgo S
   27. Lindsay S
   28. Khaitovich P
   29. Huber W
   30. Baker J
   31. Anders S
   32. Zhang YE
   33. Kaessmann H

   (2019) Gene expression across mammalian organ development

   Nature 571:505–509.

   https://doi.org/10.1038/s41586-019-1338-5

   • PubMed
   • Google Scholar
10. Book

    1. Charif D
    2. Lobry JR

    (2007)

    Seqinr 1.0-2: A contributed package to the R project for statistical computing devoted to biological sequences retrieval and analysis

    In: Bastolla U, Porto M, Roman HE, Vendruscolo M, editors. Structural Approaches to Sequence Evolution: Molecules, Networks, Populations, Biological and Medical Physics, Biomedical Engineering. Springer. pp. 207–232.

    • Google Scholar
11. 1. Chen L
    2. Bush SJ
    3. Tovar-Corona JM
    4. Castillo-Morales A
    5. Urrutia AO

    (2014) Correcting for differential transcript coverage reveals a strong relationship between alternative splicing and organism complexity

    Molecular Biology and Evolution 31:1402–1413.

    https://doi.org/10.1093/molbev/msu083

    • PubMed
    • Google Scholar
12. 1. Dutheil J
    2. Boussau B

    (2008) Non-homogeneous models of sequence evolution in the Bio++ suite of libraries and programs

    BMC Evolutionary Biology 8:255.

    https://doi.org/10.1186/1471-2148-8-255

    • PubMed
    • Google Scholar
13. 1. Dutheil JY
    2. Galtier N
    3. Romiguier J
    4. Douzery EJP
    5. Ranwez V
    6. Boussau B

    (2012) Efficient selection of branch-specific models of sequence evolution

    Molecular Biology and Evolution 29:1861–1874.

    https://doi.org/10.1093/molbev/mss059

    • PubMed
    • Google Scholar
14. 1. Figuet E
    2. Nabholz B
    3. Bonneau M
    4. Mas Carrio E
    5. Nadachowska-Brzyska K
    6. Ellegren H
    7. Galtier N

    (2016) Life history traits, protein evolution, and the nearly neutral theory in amniotes

    Molecular Biology and Evolution 33:1517–1527.

    https://doi.org/10.1093/molbev/msw033

    • PubMed
    • Google Scholar
15. 1. Freckleton RP
    2. Harvey PH
    3. Pagel M

    (2002) Phylogenetic analysis and comparative data: A test and review of evidence

    The American Naturalist 160:712–726.

    https://doi.org/10.1086/343873

    • PubMed
    • Google Scholar
16. 1. Gonzàlez-Porta M
    2. Frankish A
    3. Rung J
    4. Harrow J
    5. Brazma A

    (2013) Transcriptome analysis of human tissues and cell lines reveals one dominant transcript per gene

    Genome Biology 14:R70.

    https://doi.org/10.1186/gb-2013-14-7-r70

    • Google Scholar
17. 1. Gout JF
    2. Thomas WK
    3. Smith Z
    4. Okamoto K
    5. Lynch M

    (2013) Large-scale detection of in vivo transcription errors

    PNAS 110:18584–18589.

    https://doi.org/10.1073/pnas.1309843110

    • PubMed
    • Google Scholar
18. 1. Graveley BR

    (2001) Alternative splicing: increasing diversity in the proteomic world

    Trends in Genetics 17:100–107.

    https://doi.org/10.1016/s0168-9525(00)02176-4

    • PubMed
    • Google Scholar
19. 1. Guéguen L
    2. Gaillard S
    3. Boussau B
    4. Gouy M
    5. Groussin M
    6. Rochette NC
    7. Bigot T
    8. Fournier D
    9. Pouyet F
    10. Cahais V
    11. Bernard A
    12. Scornavacca C
    13. Nabholz B
    14. Haudry A
    15. Dachary L
    16. Galtier N
    17. Belkhir K
    18. Dutheil JY

    (2013) Bio++: efficient extensible libraries and tools for computational molecular evolution

    Molecular Biology and Evolution 30:1745–1750.

    https://doi.org/10.1093/molbev/mst097

    • PubMed
    • Google Scholar
20. 1. Guéguen L
    2. Duret L

    (2018) Unbiased estimate of synonymous and nonsynonymous substitution rates with nonstationary base composition

    Molecular Biology and Evolution 35:734–742.

    https://doi.org/10.1093/molbev/msx308

    • PubMed
    • Google Scholar
21. 1. Hamid FM
    2. Makeyev EV

    (2014) Emerging functions of alternative splicing coupled with nonsense-mediated decay

    Biochemical Society Transactions 42:1168–1173.

    https://doi.org/10.1042/BST20140066

    • PubMed
    • Google Scholar
22. 1. Hinrichs AS
    2. Karolchik D
    3. Baertsch R
    4. Barber GP
    5. Bejerano G
    6. Clawson H
    7. Diekhans M
    8. Furey TS
    9. Harte RA
    10. Hsu F
    11. Hillman-Jackson J
    12. Kuhn RM
    13. Pedersen JS
    14. Pohl A
    15. Raney BJ
    16. Rosenbloom KR
    17. Siepel A
    18. Smith KE
    19. Sugnet CW
    20. Sultan-Qurraie A
    21. Thomas DJ
    22. Trumbower H
    23. Weber RJ
    24. Weirauch M
    25. Zweig AS
    26. Haussler D
    27. Kent WJ

    (2006) The UCSC Genome Browser Database: update 2006

    Nucleic Acids Research 34:D590–D598.

    https://doi.org/10.1093/nar/gkj144

    • PubMed
    • Google Scholar
23. 1. Hsu SN
    2. Hertel KJ

    (2009) Spliceosomes walk the line: splicing errors and their impact on cellular function

    RNA Biology 6:526–530.

    https://doi.org/10.4161/rna.6.5.9860

    • PubMed
    • Google Scholar
24. 1. Huang W
    2. Massouras A
    3. Inoue Y
    4. Peiffer J
    5. Ràmia M
    6. Tarone AM
    7. Turlapati L
    8. Zichner T
    9. Zhu D
    10. Lyman RF
    11. Magwire MM
    12. Blankenburg K
    13. Carbone MA
    14. Chang K
    15. Ellis LL
    16. Fernandez S
    17. Han Y
    18. Highnam G
    19. Hjelmen CE
    20. Jack JR
    21. Javaid M
    22. Jayaseelan J
    23. Kalra D
    24. Lee S
    25. Lewis L
    26. Munidasa M
    27. Ongeri F
    28. Patel S
    29. Perales L
    30. Perez A
    31. Pu L
    32. Rollmann SM
    33. Ruth R
    34. Saada N
    35. Warner C
    36. Williams A
    37. Wu Y-Q
    38. Yamamoto A
    39. Zhang Y
    40. Zhu Y
    41. Anholt RRH
    42. Korbel JO
    43. Mittelman D
    44. Muzny DM
    45. Gibbs RA
    46. Barbadilla A
    47. Johnston JS
    48. Stone EA
    49. Richards S
    50. Deplancke B
    51. Mackay TFC

    (2014) Natural variation in genome architecture among 205 Drosophila melanogaster Genetic Reference Panel lines

    Genome Research 24:1193–1208.

    https://doi.org/10.1101/gr.171546.113

    • Google Scholar
25. 1. John S
    2. Olas JJ
    3. Mueller-Roeber B

    (2021) Regulation of alternative splicing in response to temperature variation in plants

    Journal of Experimental Botany 72:6150–6163.

    https://doi.org/10.1093/jxb/erab232

    • Google Scholar
26. 1. Kim D
    2. Paggi JM
    3. Park C
    4. Bennett C
    5. Salzberg SL

    (2019) Graph-based genome alignment and genotyping with HISAT2 and HISAT-genotype

    Nature Biotechnology 37:907–915.

    https://doi.org/10.1038/s41587-019-0201-4

    • Google Scholar
27. 1. Kimura M
    2. Maruyama T
    3. Crow JF

    (1963) The mutation load in small populations

    Genetics 48:1303–1312.

    https://doi.org/10.1093/genetics/48.10.1303

    • PubMed
    • Google Scholar
28. 1. Kozlov AM
    2. Darriba D
    3. Flouri T
    4. Morel B
    5. Stamatakis A

    (2019) RAxML-NG: a fast, scalable and user-friendly tool for maximum likelihood phylogenetic inference

    Bioinformatics 35:4453–4455.

    https://doi.org/10.1093/bioinformatics/btz305

    • PubMed
    • Google Scholar
29. 1. Kryazhimskiy S
    2. Plotkin JB

    (2008) The population genetics of dN/dS

    PLOS Genetics 4:e1000304.

    https://doi.org/10.1371/journal.pgen.1000304

    • PubMed
    • Google Scholar
30. 1. Leinonen R
    2. Sugawara H
    3. Shumway M
    4. International Nucleotide Sequence Database Collaboration

    (2011) The sequence read archive

    Nucleic Acids Research 39:D19–D21.

    https://doi.org/10.1093/nar/gkq1019

    • PubMed
    • Google Scholar
31. 1. Leung SK
    2. Jeffries AR
    3. Castanho I
    4. Jordan BT
    5. Moore K
    6. Davies JP
    7. Dempster EL
    8. Bray NJ
    9. O’Neill P
    10. Tseng E
    11. Ahmed Z
    12. Collier DA
    13. Jeffery ED
    14. Prabhakar S
    15. Schalkwyk L
    16. Jops C
    17. Gandal MJ
    18. Sheynkman GM
    19. Hannon E
    20. Mill J

    (2021) Full-length transcript sequencing of human and mouse cerebral cortex identifies widespread isoform diversity and alternative splicing

    Cell Reports 37:110022.

    https://doi.org/10.1016/j.celrep.2021.110022

    • PubMed
    • Google Scholar
32. 1. Li W
    2. Lynch M

    (2020) Universally high transcript error rates in bacteria

    eLife 9:e54898.

    https://doi.org/10.7554/eLife.54898

    • PubMed
    • Google Scholar
33. 1. Liu Z
    2. Zhang J

    (2018a) Human C-to-U Coding RNA Editing Is Largely Nonadaptive

    Molecular Biology and Evolution 35:963–969.

    https://doi.org/10.1093/molbev/msy011

    • PubMed
    • Google Scholar
34. 1. Liu Z
    2. Zhang J

    (2018b) Most m6A RNA Modifications in Protein-Coding Regions Are Evolutionarily Unconserved and Likely Nonfunctional

    Molecular Biology and Evolution 35:666–675.

    https://doi.org/10.1093/molbev/msx320

    • PubMed
    • Google Scholar
35. 1. Löytynoja A
    2. Goldman N

    (2008) Phylogeny-aware gap placement prevents errors in sequence alignment and evolutionary analysis

    Science 320:1632–1635.

    https://doi.org/10.1126/science.1158395

    • PubMed
    • Google Scholar
36. 1. Lynch M
    2. Conery JS

    (2003) The origins of genome complexity

    Science 302:1401–1404.

    https://doi.org/10.1126/science.1089370

    • PubMed
    • Google Scholar
37. 1. Lynch M

    (2006) The origins of eukaryotic gene structure

    Molecular Biology and Evolution 23:450–468.

    https://doi.org/10.1093/molbev/msj050

    • PubMed
    • Google Scholar
38. 1. Lynch M

    (2007) The frailty of adaptive hypotheses for the origins of organismal complexity

    PNAS 104:8597–8604.

    https://doi.org/10.1073/pnas.0702207104

    • Google Scholar
39. 1. Lynch M
    2. Ackerman MS
    3. Gout JF
    4. Long H
    5. Sung W
    6. Thomas WK
    7. Foster PL

    (2016) Genetic drift, selection and the evolution of the mutation rate

    Nature Reviews. Genetics 17:704–714.

    https://doi.org/10.1038/nrg.2016.104

    • PubMed
    • Google Scholar
40. 1. Mackay TFC
    2. Richards S
    3. Stone EA
    4. Barbadilla A
    5. Ayroles JF
    6. Zhu D
    7. Casillas S
    8. Han Y
    9. Magwire MM
    10. Cridland JM
    11. Richardson MF
    12. Anholt RRH
    13. Barrón M
    14. Bess C
    15. Blankenburg KP
    16. Carbone MA
    17. Castellano D
    18. Chaboub L
    19. Duncan L
    20. Harris Z
    21. Javaid M
    22. Jayaseelan JC
    23. Jhangiani SN
    24. Jordan KW
    25. Lara F
    26. Lawrence F
    27. Lee SL
    28. Librado P
    29. Linheiro RS
    30. Lyman RF
    31. Mackey AJ
    32. Munidasa M
    33. Muzny DM
    34. Nazareth L
    35. Newsham I
    36. Perales L
    37. Pu L-L
    38. Qu C
    39. Ràmia M
    40. Reid JG
    41. Rollmann SM
    42. Rozas J
    43. Saada N
    44. Turlapati L
    45. Worley KC
    46. Wu Y-Q
    47. Yamamoto A
    48. Zhu Y
    49. Bergman CM
    50. Thornton KR
    51. Mittelman D
    52. Gibbs RA

    (2012) The Drosophila melanogaster genetic reference panel

    Nature 482:173–178.

    https://doi.org/10.1038/nature10811

    • Google Scholar
41. 1. Mazin PV
    2. Khaitovich P
    3. Cardoso-Moreira M
    4. Kaessmann H

    (2021) Alternative splicing during mammalian organ development

    Nature Genetics 53:925–934.

    https://doi.org/10.1038/s41588-021-00851-w

    • Google Scholar
42. 1. McGlincy NJ
    2. Smith CWJ

    (2008) Alternative splicing resulting in nonsense-mediated mRNA decay: what is the meaning of nonsense?

    Trends in Biochemical Sciences 33:385–393.

    https://doi.org/10.1016/j.tibs.2008.06.001

    • PubMed
    • Google Scholar
43. 1. Merkin J
    2. Russell C
    3. Chen P
    4. Burge CB

    (2012) Evolutionary dynamics of gene and isoform regulation in Mammalian tissues

    Science 338:1593–1599.

    https://doi.org/10.1126/science.1228186

    • PubMed
    • Google Scholar
44. 1. Mudge JM
    2. Frankish A
    3. Fernandez-Banet J
    4. Alioto T
    5. Derrien T
    6. Howald C
    7. Reymond A
    8. Guigó R
    9. Hubbard T
    10. Harrow J

    (2011) The origins, evolution, and functional potential of alternative splicing in vertebrates

    Molecular Biology and Evolution 28:2949–2959.

    https://doi.org/10.1093/molbev/msr127

    • PubMed
    • Google Scholar
45. 1. Ohta T

    (1973) Slightly deleterious mutant substitutions in evolution

    Nature 246:96–98.

    https://doi.org/10.1038/246096a0

    • PubMed
    • Google Scholar
46. 1. Pickrell JK
    2. Pai AA
    3. Gilad Y
    4. Pritchard JK

    (2010) Noisy Splicing Drives mRNA Isoform Diversity in Human Cells

    PLOS Genetics 6:e1001236.

    https://doi.org/10.1371/journal.pgen.1001236

    • Google Scholar
47. 1. Rajon E
    2. Masel J

    (2011) Evolution of molecular error rates and the consequences for evolvability

    PNAS 108:1082–1087.

    https://doi.org/10.1073/pnas.1012918108

    • PubMed
    • Google Scholar
48. 1. Reyes A
    2. Anders S
    3. Weatheritt RJ
    4. Gibson TJ
    5. Steinmetz LM
    6. Huber W

    (2013) Drift and conservation of differential exon usage across tissues in primate species

    PNAS 110:15377–15382.

    https://doi.org/10.1073/pnas.1307202110

    • PubMed
    • Google Scholar
49. 1. Roberts A
    2. Pimentel H
    3. Trapnell C
    4. Pachter L

    (2011) Identification of novel transcripts in annotated genomes using RNA-Seq

    Bioinformatics 27:2325–2329.

    https://doi.org/10.1093/bioinformatics/btr355

    • PubMed
    • Google Scholar
50. Preprint

    1. Romiguier J
    2. Weyna A

    (2020) Relaxation of Purifying Selection Suggests Low Effective Population Size in Eusocial Hymenoptera and Solitary Pollinating Bees

    bioRxiv.

    https://doi.org/10.1101/2020.04.14.038893

    • Google Scholar
51. 1. Saudemont B
    2. Popa A
    3. Parmley JL
    4. Rocher V
    5. Blugeon C
    6. Necsulea A
    7. Meyer E
    8. Duret L

    (2017) The fitness cost of mis-splicing is the main determinant of alternative splicing patterns

    Genome Biology 18:208.

    https://doi.org/10.1186/s13059-017-1344-6

    • PubMed
    • Google Scholar
52. 1. Seppey M
    2. Manni M
    3. Zdobnov EM

    (2019) BUSCO: Assessing Genome Assembly and Annotation Completeness

    Methods in Molecular Biology 1962:227–245.

    https://doi.org/10.1007/978-1-4939-9173-0_14

    • PubMed
    • Google Scholar
53. 1. Singh P
    2. Ahi EP

    (2022) The importance of alternative splicing in adaptive evolution

    Molecular Ecology 31:1928–1938.

    https://doi.org/10.1111/mec.16377

    • PubMed
    • Google Scholar
54. 1. Tomso DJ
    2. Bell DA

    (2003) Sequence context at human single nucleotide polymorphisms: overrepresentation of CpG dinucleotide at polymorphic sites and suppression of variation in CpG islands

    Journal of Molecular Biology 327:303–308.

    https://doi.org/10.1016/s0022-2836(03)00120-7

    • PubMed
    • Google Scholar
55. 1. Traverse CC
    2. Ochman H

    (2016) Conserved rates and patterns of transcription errors across bacterial growth states and lifestyles

    PNAS 113:3311–3316.

    https://doi.org/10.1073/pnas.1525329113

    • PubMed
    • Google Scholar
56. 1. Tress ML
    2. Abascal F
    3. Valencia A

    (2017a) Alternative splicing may not be the key to proteome complexity

    Trends in Biochemical Sciences 42:98–110.

    https://doi.org/10.1016/j.tibs.2016.08.008

    • Google Scholar
57. 1. Tress ML
    2. Abascal F
    3. Valencia A

    (2017b) Most alternative isoforms are not functionally important

    Trends in Biochemical Sciences 42:408–410.

    https://doi.org/10.1016/j.tibs.2017.04.002

    • Google Scholar
58. 1. Verta JP
    2. Jacobs A

    (2022) The role of alternative splicing in adaptation and evolution

    Trends in Ecology & Evolution 37:299–308.

    https://doi.org/10.1016/j.tree.2021.11.010

    • Google Scholar
59. 1. Waples RS

    (2016) Life-history traits and effective population size in species with overlapping generations revisited: the importance of adult mortality

    Heredity 117:241–250.

    https://doi.org/10.1038/hdy.2016.29

    • PubMed
    • Google Scholar
60. 1. Wright CJ
    2. Smith CWJ
    3. Jiggins CD

    (2022) Alternative splicing as a source of phenotypic diversity

    Nature Reviews Genetics 23:697–710.

    https://doi.org/10.1038/s41576-022-00514-4

    • Google Scholar
61. 1. Xiong K
    2. McEntee JP
    3. Porfirio DJ
    4. Masel J

    (2017) Drift barriers to quality control when genes are expressed at different levels

    Genetics 205:397–407.

    https://doi.org/10.1534/genetics.116.192567

    • Google Scholar
62. 1. Xu G
    2. Zhang J

    (2014) Human coding RNA editing is generally nonadaptive

    PNAS 111:3769–3774.

    https://doi.org/10.1073/pnas.1321745111

    • PubMed
    • Google Scholar
63. 1. Xu C
    2. Zhang J

    (2018) Alternative polyadenylation of mammalian transcripts is generally deleterious, not adaptive

    Cell Systems 6:734–742.

    https://doi.org/10.1016/j.cels.2018.05.007

    • PubMed
    • Google Scholar
64. 1. Xu C
    2. Park J-K
    3. Zhang J

    (2019) Evidence that alternative transcriptional initiation is largely nonadaptive

    PLOS Biology 17:e3000197.

    https://doi.org/10.1371/journal.pbio.3000197

    • PubMed
    • Google Scholar
65. 1. Xu C
    2. Zhang J

    (2020) A different perspective on alternative cleavage and polyadenylation

    Nature Reviews. Genetics 21:63.

    https://doi.org/10.1038/s41576-019-0198-z

    • PubMed
    • Google Scholar
66. 1. Yang Z
    2. Nielsen R

    (1998) Synonymous and nonsynonymous rate variation in nuclear genes of mammals

    Journal of Molecular Evolution 46:409–418.

    https://doi.org/10.1007/pl00006320

    • PubMed
    • Google Scholar
67. 1. Zhang J
    2. Xu C

    (2022) Gene product diversity: adaptive or not?

    Trends in Genetics 38:1112–1122.

    https://doi.org/10.1016/j.tig.2022.05.002

    • PubMed
    • Google Scholar

Author details

1. Florian Bénitière

   Laboratoire de Biometrie et Biologie Evolutive, CNRS, Universite Lyon 1, Villeurbanne, France

   Contribution

   Resources, Data curation, Software, Formal analysis, Investigation, Methodology, Writing - original draft, Writing - review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-7773-3542

2. Anamaria Necsulea

   Laboratoire de Biometrie et Biologie Evolutive, CNRS, Universite Lyon 1, Villeurbanne, France

   Contribution

   Conceptualization, Resources, Data curation, Software, Formal analysis, Supervision, Funding acquisition, Investigation, Methodology, Writing - original draft, Project administration, Writing - review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-9861-7698

3. Laurent Duret

   Laboratoire de Biometrie et Biologie Evolutive, CNRS, Universite Lyon 1, Villeurbanne, France

   Contribution

   Conceptualization, Formal analysis, Supervision, Funding acquisition, Investigation, Methodology, Writing - original draft, Project administration, Writing - review and editing

   For correspondence

   Laurent.Duret@univ-lyon1.fr

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-2836-3463

• 1,459

  views

• 98

  downloads

• 1

  citations

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