Search

The coalescent model represents how individuals sampled from a population may have originated from a last common ancestor. The bounded coalescent model is obtained by conditioning the coalescent model such that the last common ancestor must have existed after a certain date. This conditioned model arises in a variety of applications, such as speciation, horizontal gene transfer or transmission analysis, and yet the bounded coalescent model has not been previously analysed in detail. Here we describe a new algorithm to simulate from this model directly, without resorting to rejection sampling. We show that this direct simulation algorithm is more computationally efficient than the rejection sampling approach. We also show how to calculate the probability of the last common ancestor occurring after a given date, which is required to compute the probability of realisations under the bounded coalescent model. Our results are applicable in both the isochronous (when all samples have the same date) and heterochronous (where samples can have different dates) settings. We explore the effect of setting a bound on the date of the last common ancestor, and show that it affects a number of properties of the resulting phylogenies. All our methods are implemented in a new R package called BoundedCoalescent which is freely available online.

A hospital outbreak of carbapenem-resistant Enterobacterales was detected by routine surveillance. Whole genome sequencing and subsequent analysis revealed a conserved promiscuous

The incidence of carbapenemase-producing Enterobacterales (CPE) is rising globally, yet Imipenemase (IMP) carbapenemases remain relatively rare. This study describes an investigation of the emergence of IMP-encoding CPE amongst diverse Enterobacterales species between 2016 and 2019 in patients across a London regional hospital network.A network analysis approach to patient pathways, using routinely collected electronic health records, identified previously unrecognised contacts between patients who were IMP CPE positive on screening, implying potential bacterial transmission events. Whole genome sequencing of 85 Enterobacterales isolates from these patients revealed that 86% (73/85) were diverse species (predominantly Klebsiella spp, Enterobacter spp, E. coli) and harboured an IncHI2 plasmid, which carried both blaIMP and the putative mobile colistin resistance gene mcr-9. Detailed phylogenetic analysis identified two distinct IncHI2 plasmid lineages, A and B, both of which showed significant association with patient movements between four hospital sites and across medical specialities.Combined, our patient network and plasmid analyses demonstrate an interspecies, plasmid-mediated outbreak of blaIMPCPE, which remained unidentified during standard microbiology and infection control investigations. With whole genome sequencing (WGS) technologies and large-data incorporation, the outbreak investigation approach proposed here provides a framework for real-time identification of key factors causing pathogen spread. Analysing outbreaks at the plasmid level reveals that resistance may be wider spread than suspected, allowing more targetted interventions to stop the transmission of resistance within hospital networks.

A hospital outbreak of carbapenem-resistant Enterobacterales was detected by routine surveillance. Whole genome sequencing and subsequent analysis revealed a conserved promiscuous bla OXA-48 carrying plasmid as the defining factor within this outbreak. Four different species of Enterobacterales were involved in the outbreak. Escherichia coli ST399 accounted for 35 of all the 55 isolates. Comparative genomics analysis using publicly available E. coli ST399 genomes showed that the outbreak E. coli ST399 isolates formed a unique clade. We developed a mathematical model of pOXA-48-like plasmid transmission between host lineages and used it to estimate its conjugation rate, giving a lower bound of 0.23 conjugation events per lineage per year. Our analysis suggests that co-evolution between the pOXA-48-like plasmid and E. coli ST399 could have played a role in the outbreak. This is the first study to report carbapenem-resistant E. coli ST399 carrying blaOXA-48 as the main cause of a plasmid-borne outbreak within a hospital setting. Our findings suggest complementary roles for both plasmid conjugation and clonal expansion in the emergence of this outbreak.

Microbial population genetics models often assume that all lineages are constrained by the same population size dynamics over time. However, many neutral and selective events can invalidate this assumption and can contribute to the clonal expansion of a specific lineage relative to the rest of the population. Such differential phylodynamic properties between lineages result in asymmetries and imbalances in phylogenetic trees that are sometimes described informally but which are difficult to analyze formally. To this end, we developed a model of how clonal expansions occur and affect the branching patterns of a phylogeny. We show how the parameters of this model can be inferred from a given dated phylogeny using Bayesian statistics, which allows us to assess the probability that one or more clonal expansion events occurred. For each putative clonal expansion event, we estimate its date of emergence and subsequent phylodynamic trajectory, including its long-term evolutionary potential which is important to determine how much effort should be placed on specific control measures. We demonstrate the applicability of our methodology on simulated and real data sets. Inference under our clonal expansion model can reveal important features in the evolution and epidemiology of infectious disease pathogens. [Clonal expansion; genomic epidemiology; microbial population genomics; phylodynamics.]

Introduction Many countries with an early outbreak of SARS-CoV-2 struggled to gauge the size and start date of the epidemic mainly due to limited testing capacities and a large proportion of undetected asymptomatic and mild infections. Iran was among the first countries with a major outbreak outside China. Methods We constructed a globally representative sample of 802 genomes, including 46 samples from patients inside or with a travel history to Iran. We then performed a phylogenetic analysis to identify clades related to samples from Iran and estimated the start of the epidemic and early doubling times in cases. We leveraged air travel data from 36 exported cases of COVID-19 to estimate the point-prevalence and the basic reproductive number across the country. We also analysed the province-level all-cause mortality data during winter and spring 2020 to estimate under-reporting of COVID-19-related deaths. Finally, we use this information in an SEIR model to reconstruct the early outbreak dynamics and assess the effectiveness of intervention measures in Iran. Results By identifying the most basal clade that contained genomes from Iran, our phylogenetic analysis showed that the age of the root is placed on 2019-12-21 (95 % HPD: 2019-09-07 – 2020-02-14). This date coincides with our estimated epidemic start date on 2019-12-25 (95 %CI: 2019-12-11 – 2020-02-24) based air travel data from exported cases with an early doubling time of 4.0 (95 %CI: 1.4–6.7) days in cases. Our analysis of all-cause mortality showed 21.9 (95 % CI: 16.7–27.2) thousand excess deaths by the end of summer. Our model forecasted the second epidemic peak and suggested that by 2020-08-31 a total of 15.0 (95 %CI: 4.9–25.0) million individuals recovered from the disease across the country. Conclusion These findings have profound implications for assessing the stage of the epidemic in Iran despite significant levels of under-reporting. Moreover, the results shed light on the dynamics of SARS-CoV-2 transmissions in Iran and central Asia. They also suggest that in the absence of border screening, there is a high risk of introduction from travellers from areas with active outbreaks. Finally, they show both that well-informed epidemic models are able to forecast episodes of resurgence following a relaxation of interventions, and that NPIs are key to controlling ongoing epidemics.

Highlights • COVID-19 measures were applied at similar dates across provinces in China • Hubei showed much greater disease severity compared to other provinces • Provincial data on epidemics and interventions are available for further research, Objectives This data collation effort aims to provide a comprehensive database to describe the epidemic trends and responses during the first wave of coronavirus disease 2019 (COVID-19) across main provinces in China. Methods From mid-January to March 2020, we extracted publicly available data on the spread and control of COVID-19 from 31 provincial health authorities and major media outlets in mainland China. Based on these data, we conducted a descriptive analysis of the epidemics in the six most-affected provinces. Results School closures, travel restrictions, community-level lockdown, and contact tracing were introduced concurrently around late January but subsequent epidemic trends were different across provinces. Compared to Hubei, the other five most-affected provinces reported a lower crude case fatality ratio and proportion of critical and severe hospitalised cases. From March 2020, as local transmission of COVID-19 declined, switching the focus of measures to testing and quarantine of inbound travellers could help to sustain the control of the epidemic. Conclusions Aggregated indicators of case notifications and severity distributions are essential for monitoring an epidemic. A publicly available database with these indicators and information on control measures provides a useful source for exploring further research and policy planning for response to the COVID-19 epidemic.

The expression of accessory non-structural proteins V and W in Newcastle disease virus (NDV) infections depends on RNA editing. These proteins are derived from frameshifts of the sequence coding for the P protein via co-transcriptional insertion of one or two guanines in the mRNA. However, a larger number of guanines can be inserted with lower frequencies. We analysed data from deep RNA sequencing of samples from in vitro and in vivo NDV infections to uncover the patterns of mRNA editing in NDV. The distribution of insertions is well described by a simple Markov model of polymerase stuttering, providing strong quantitative confirmation of the molecular process hypothesised by Kolakofsky and collaborators three decades ago. Our results suggest that the probability that the NDV polymerase would stutter is about 0.45 initially, and 0.3 for further subsequent insertions. The latter probability is approximately independent of the number of previous insertions, the host cell, and viral strain. However, in LaSota infections, we also observe deviations from the predicted V/W ratio of about 3:1 according to this model, which could be attributed to deviations from this stuttering model or to further mechanisms downregulating the abundance of W protein.

Many countries with an early outbreak of SARS-CoV-2 struggled to gauge the size and start date of the epidemic mainly due to limited testing capacities and a large proportion of undetected asymptomatic and mild infections. Iran was among the first countries with a major outbreak outside China. Using all genomic sequences collected from patients with a travel link to Iran, we estimate that the epidemic started on 21/01/2020 (95% HPD: 05/12/2019 – 14/02/2020) with a doubling time of 3 days (95% HPD: 1.68 – 16.27). We also show, using air travel data from confirmed exported cases, that from late February to early March the number of active cases across the country were more than a hundred times higher than the reported cases at the time. A detailed province-level analysis of all-cause mortality shows 20,718 (CI 95%: 18,859 – 22,576) excess deaths during winter and spring 2020 compared to previous years, almost twice the number of reported COVID-19-related deaths at the time. Correcting for under-reporting of prevalence and deaths, we use an SEIR model to reconstruct the outbreak dynamics in Iran. Our model forecasted the second epidemic peak and suggests that by 14/07/2020 a total of 9M (CI 95%: 118K – 44M) have recovered from the disease across the country. These findings have profound implications for assessing the stage of the epidemic in Iran and shed light on the dynamics of SARS-CoV-2 transmissions in Iran and central Asia despite significant levels of under-reporting.

A hospital outbreak of carbapenem-resistant Enterobacteriales was detected by routine surveillance. Whole genome sequencing and subsequent analysis revealed a conserved promiscuous OXA-48 carrying plasmid as the defining factor within this outbreak. Four different species of Enterobacteriales were involved in the outbreak.Escherichia coliST399 accounted for 20/55 of all the isolates. Comparative genomics with publicly availableE. coliST399 sequence data showed that the outbreak isolates formed a unique clade. The OXA-48 plasmid identified in the outbreak differed from other known plasmids by an estimated five homologous recombination events. We estimated a lower bound to the plasmid conjugation rate to be 0.23 conjugation events per lineage per year. Our analysis suggests co-evolution between the plasmid and its main bacterial host to be a key driver of the outbreak. This is the first study to report carbapenem-resistantE. coliST399 carrying OXA48 as the main cause of a plasmid-borne outbreak within a hospital setting. This study supports complementary roles for both plasmid conjugation and clonal expansion in the emergence of this outbreak.

Since the first outbreak in China, the Coronavirus Disease 2019 (COVID-19) has rapidly spread around the world. Iran was one of the first countries outside of China to report infections with COVID-19. With nearly 100 exported cases to various other countries, it has since been the epicentre of the outbreak in the Middle east. By examining the age-stratified COVID-19 case fatality rates across the country and 14 university hospitals in Tehran, we find that, in younger age groups, the reported cases on 13/03/2020 only capture less than 10% of symptomatic cases in the population. This indicates significant levels of under-reporting in Iran. Using the 18 full-genome sequences from cases with a travel history or link to Iran, as well as the one full genome sequence obtained from within the country, we estimate the time to the most recent common ancestor of sequences which suggests the likely start of the outbreak on 21/01/2020 (95% HPD: 05/12/2019 - 14/02/2020) with an approximate doubling time of 3.07 (95% HPD: 1.68 - 16.27). Also, based on known exported cases to Oman, Kuwait, Lebanon, and China, we estimate the outbreak size on 25 February and 6 March to be around 13,700 (95% CI: 7,600 - 33,300) and 60,500 (43,200 - 209,200), respectively. Knowing the size of the outbreak at two time points and the typical doubling times associated with the COVID-19 epidemics in countries across Europe and North America, we can independently verify that the likely start of epidemic in Iran is around 15/01/2020 (27/12/2019 - 24/01/2020). Our assessment of the fate of the epidemic based on current levels of non-pharmaceutical interventions implemented by the government suggests upward of 10 million cases (IQR: 6.7×106 - 1.8×107) and 100,000 ICU beds required (IQR: 7.7 × 104 - 1.4 × 105) during the peak of the epidemic with more than 100,000 cumulative deaths (IQR: 1.8 × 105 - 2.4 × 105). We also predict a peak in demand for ICU beds on 21/04/2020 (IQR: 06/04/2020 - 23/05/2020). The large span of the peak of the ICU demand is a result of two separate peaks, with the first occurring at around 15/4/2020 and the second in approximately a months time. The latter is also expected to last longer and is based on the relatively relaxed social distancing measures in place. The exact magnitude and timing of the peaks strictly depends on levels of interventions and can change significantly upon new information or change of policy. We caution that a lack of, or relaxed, stringent intervention measures, during a period of highly under-reported spread, would likely lead to the healthcare system becoming overwhelmed in the next few months.

PurposeThe COVID-19 death-rate in Italy continues to climb, surpassing that in every other country. We implement one of the first nationally representative surveys about this unprecedented public health crisis and use it to evaluate the Italian government’ public health efforts and citizen responses.Findings(1) Public health messaging is being heard. Except for slightly lower compliance among young adults, all subgroups we studied understand how to keep themselves and others safe from the SARS-Cov-2 virus. Remarkably, even those who do not trust the government, or think the government has been untruthful about the crisis believe the messaging and claim to be acting in accordance. (2) The quarantine is beginning to have serious negative effects on the population’s mental health.Policy RecommendationsCommunications should move from explaining to citizens that they should stay at home to what they can do there. We need interventions that make staying following public health protocols more desirable, such as virtual social interactions, online social reading activities, classes, exercise routines, etc. — all designed to reduce the boredom of long term social isolation and to increase the attractiveness of following public health recommendations. Interventions like these will grow in importance as the crisis wears on around the world, and staying inside wears on people.

Background: The timing of SARS-CoV-2 transmission is a critical factor to understand the epidemic trajectory and the impact of isolation, contact tracing and other non-pharmaceutical interventions on the spread of COVID-19 epidemics. Methods: We examined the distribution of transmission event times with respect to exposure and onset of symptoms. We analysed 119 transmission pairs with known date of onset of symptoms for both index and secondary cases and partial information on their intervals of exposure. We inferred the distribution for generation time and time from onset of symptoms to transmission by maximum likelihood. We modelled different relations between time of infection, onset of symptoms and transmission, inferring the most appropriate one according to the Akaike Information Criterion. Finally, we estimated the fraction of pre-symptomatic and early symptomatic transmissions among all pairs using a Bayesian approach. Findings: For symptomatic individuals, the timing of transmission of SARS-CoV-2 was more directly linked to the onset of clinical symptoms of COVID-19 than to the time since infection. The time of transmission was approximately centered and symmetric around the onset of symptoms, with three quarters of events occurring in the window from 2-3 days before to 2-3 days after. The pre-symptomatic infectious period extended further back in time for individuals with longer incubation periods. Overall, the fraction of transmission from strictly pre-symptomatic infections was high (41%; 95%CI 31-50%), but a comparably large fraction of transmissions occurred on the same day as the onset of symptoms or the next day (35%; 95%CI 26-45%). We caution against overinterpretation of the fraction and timing of late symptomatic transmissions, due to their dependence on behavioural factors and interventions. Interpretation: Infectiousness is causally driven by the onset of symptoms. Public health authorities should reassess their policies on the contact tracing window in the light of individual variability in presymptomatic infectious period. Information about when a case was infected should be collected where possible, in order to assess how far into the past their contacts should be traced. The large fraction of transmission from strictly pre-symptomatic infections limits the efficacy of symptom-based interventions, while the large fraction of early symptomatic transmissions underlines the critical importance of individuals distancing themselves from others as soon as they notice any symptoms, even if mild. Rapid or at-home testing and contextual risk information could greatly facilitate efficient early isolation. Funding Statement: The study was funded by an award from the Li Ka Shing Foundation to CF. Declaration of Interests: None of the authors have competing financial or non-financial interests.

We undertook a comprehensive comparative analysis of a collection of 30 small (

One of the major challenges in evolutionary biology is the identification of the genetic basis of postzygotic reproductive isolation. Given its pivotal role in this process, here we explore the drivers that may account for the evolutionary dynamics of the PRDM9 gene between continental and island systems of chromosomal variation in house mice. Using a data set of nearly 400 wild-caught mice of Robertsonian systems, we identify the extent of PRDM9 diversity in natural house mouse populations, determine the phylogeography of PRDM9 at a local and global scale based on a new measure of pairwise genetic divergence, and analyze selective constraints. We find 57 newly described PRDM9 variants, this diversity being especially high on Madeira Island, a result that is contrary to the expectations of reduced variation for island populations. Our analysis suggest that the PRDM9 allelic variability observed in Madeira mice might be influenced by the presence of distinct chromosomal fusions resulting from a complex pattern of introgression or multiple colonization events onto the island. Importantly, we detect a significant reduction in the proportion of PRDM9 heterozygotes in Robertsonian mice, which showed a high degree of similarity in the amino acids responsible for protein–DNA binding. Our results suggest that despite the rapid evolution of PRDM9 and the variability detected in natural populations, functional constraints could facilitate the accumulation of allelic combinations that maintain recombination hotspot symmetry. We anticipate that our study will provide the basis for examining the role of different PRDM9 genetic backgrounds in reproductive isolation in natural populations.

Objectives Methicil l in - resistant Staphylococcus aureus (MRSA) is a leading cause of hospital - associated infection. Acquired r esistance is encoded by the mecA gene or its homologue mecC but little is known about the evolutionary dynamics involved in gain and loss of resistance. The objective of this study was to obtain an expanded understanding of S. aureus methicilin resistance micro evolution in vivo , by focusing on a single lineage . Methods We compared the whole genome sequences of 231 isolates from a single epidemic lineage (clonal complex CC30 and spa - type t018) of S. aureus that caused an epidemic in the United Kingdom. Results We show that resistance to methicillin in this single lineage was gained on at least two separate occasions , one of which led to a clonal expansion around 1995 presumably caused by a selective advantage . Resistance was however subsequently lost in vivo by nine strains isolated between 2008 and 201 2 . We describe the genetic mechanisms involved in this loss of 40 resistance and the imperfect relationship between genotypic and phenotypic resistance . Conclusions The recent re - emergence of methicillin susceptibility in this epidemic lineage suggests a significant fitness cost of resistance and reduced selective advantage following the introduction in the mid 2000s of MRSA hospital control measures throughout the United Kingdom.

bioRxiv preprint, We investigate the dependence of the site frequency spectrum on the topological structure of genealogical trees. We show that basic population genetic statistics, for instance, estimators of θ or neutrality tests such as Tajima’s D, can be decomposed into components of waiting times between coalescent events and of tree topology. Our results clarify the relative impact of the two components on these statistics. We provide a rigorous interpretation of positive or negative values of an important class of neutrality tests in terms of the underlying tree shape. In particular, we show that values of Tajima’s D and Fay and Wu’s H depend in a direct way on a peculiar measure of tree balance, which is mostly determined by the root balance of the tree. We present a new test for selection in the same class as Fay and Wu’s H and discuss its interpretation and power. Finally, we determine the trees corresponding to extreme expected values of these neutrality tests and present formulas for these extreme values as a function of sample size and number of segregating sites., A.L. is funded by the United Kingdom National Institute for Health Research, Health Protection Research Unit on Modelling Methodology (grant HPRU-2012-10080). L.F. and G.A. acknowledge support from the grant ANR-12-JSV7-0007 from Agence Nationale de Recherche (France). G.A. acknowledges support from the grant ANR-12-BSV7-0012-04 from Agence Nationale de Recherche (France). T.W. acknowledges support from DFG-SPP1590 by the German Science Foundation. S.E.R.O. is supported by grants CGL2009-09346 (MICINN, Spain), AGL2013-41834-R (MEC, Spain), by the CERCA Programme/Generalitat de Catalunya and acknowledges financial support from the Spanish Ministry of Economy and Competitiveness, through the Severo Ochoa Programme for Centres of Excellence in R&D 2016-2019 (SEV‐2015‐0533).

Sequence motifs are words of nucleotides in DNA with biological functions, e.g., gene regulation. Identification of such words proceeds through rejection of Markov models on the expected motif frequency along the genome. Additional biological information can be extracted from the correlation structure among patterns of motif occurrences. In this paper a log-linear multivariate intensity Poisson model is estimated via expectation maximization on a set of motifs along the genome of E. coli K12. The proposed approach allows for excitatory as well as inhibitory interactions among motifs and between motifs and other genomic features like gene occurrences. Our findings confirm previous stylized facts about such types of interactions and shed new light on genome-maintenance functions of some particular motifs. We expect these methods to be applicable to a wider set of genomic features., L.F. acknowledges support from CSIC (Spain) under the JAE-doc program. Work was funded by grants AGL2010-14822 (MICINN, Spain) to M. Perez-Enciso and Consolider project (MICINN, Spain) to the Centre of Research in Agricultural Genomics.

International audience; The Escherichia coli species is divided in phylogenetic groups that differ in their virulence and commensal distribution. Strains belonging to the B2 group are involved in extra-intestinal pathologies but also appear to be more prevalent as commensals among human occidental populations. To investigate the genetic specificities of B2 subgroup , we used 128 sequenced genomes and identified genes of the core genome that showed marked difference between B2 and non-B2 genomes. We focused on the gene and its surrounding region with the strongest divergence between B2 and non-B2, the antiporter gene nhaA. This gene is part of the nhaAR operon, which is in the core genome but flanked by mobile regions, and is involved in growth at high pH and high sodium concentrations. Consistently, we found that a panel of non-B2 strains grew faster than B2 at high pH and high sodium concentrations. However, we could not identify differences in expression of the nhaAR operon using fluorescence reporter plasmids. Furthermore, the operon deletion had no differential impact between B2 and non-B2 strains, and did not result in a fitness modification in a murine model of gut colonization. Nevertheless, sequence analysis and experiments in a murine model of septicemia revealed that recombination in nhaA among B2 strains was observed in strains with low virulence. Finally, nhaA and nhaAR operon deletions drastically decreased virulence in one B2 strain. This effect of nhaAR deletion appeared to be stronger than deletion of all pathogenicity islands. Thus, a population genetic approach allowed us to identify an operon in the core genome without strong effect in commensalism but with an important role in extra-intestinal virulence, a landmark of the B2 strains.

Amino acid tandem repeats, also known as homopeptides, are regions within proteins characterized by the consecutive recurrence of a single amino acid. In humans, ∼15%–20% of the proteins contain at least one amino acid tandem repeat of size 5 or longer (Karlin et al. 2002; Alba and Guigo 2004). At the DNA level they might be encoded by a run of pure codons or by a mixture of synonymous codons. Runs of pure codons are more frequent than expected given the frequencies of each individual codon (Alba et al. 1999a, 2001), which reflects the influence of trinucleotide slippage in amino acid tandem repeat expansion. Due to the high mutation rates associated with slippage (Weber and Wong 1993), repeats are an important source of genetic variability (Wren et al. 2000; Mularoni et al. 2006) and may contribute to adaptive processes (Fondon and Garner 2004; Kashi and King 2006). It has also recently been shown that alternatively spliced genes are enriched in both the number and the length of amino acid tandem repeats (Haerty and Golding 2010). Not surprisingly, coding repeats often show high interspecies size variation. For example, ∼17% of the repeats in orthologous human and chimpanzee proteins differ in size in the two species (Mularoni et al. 2008). In humans, the uncontrolled expansion of CAG triplets encoding glutamine tracts is associated with a number of neurological disorders, such as Huntington disease and several ataxias (Brown and Brown 2004; Gatchel and Zoghbi 2005). Mutations generating abnormally long alanine tracts also cause several human developmental diseases (Brown and Brown 2004). The role of amino acid tandem repeats in protein function remains a matter of controversy. Repeats are generally located in protein regions that show low sequence conservation when compared to other parts of the protein (Hancock et al. 2001; Faux et al. 2007). It has been claimed that the high frequency of amino acid tandem repeats may simply reflect their high probability of expansion rather than their functional importance (Lovell 2003). In addition, the analysis of the codon composition of serine repeats has revealed that the majority of them are strongly influenced by slippage, with few cases showing clear evidence of selection (Huntley and Golding 2006). Although the vast majority of repeats do not have a known function, several examples exist in which modifications of the size of glutamine, proline, or alanine tracts can alter the capacity of the protein to activate or repress transcription (Gerber et al. 1994; Lanz et al. 1995; Janody et al. 2001; Galant and Carroll 2002; Buchanan et al. 2004; Brown et al. 2005). It has also been recently discovered that histidine repeats mediate nuclear speckle trafficking in several transcription factors and neural development proteins (Alvarez et al. 2003; Salichs et al. 2009). Finally, amino acid tandem repeats are often embedded in highly disordered protein regions (Huntley and Golding 2002; Tompa 2003; Simon and Hancock 2009), and thus may be involved in low-affinity or transient interactions that are hard to determine experimentally (Dunker et al. 2008). Studies aimed at understanding vertebrate amino acid tandem repeat evolution have mainly focused on the comparison of orthologous proteins from pairs of related species (Alba et al. 1999b; Hancock et al. 2001; Faux et al. 2007; Mularoni et al. 2007, 2008; Simon and Hancock 2009). While strongly conserved repeats tend to be encoded by a mixture of synonymous codons, nonconserved repeats are more often encoded by runs of identical codons (Alba et al. 1999b; Alba and Guigo 2004; Mularoni et al. 2007). This is consistent with a predominant role of slippage in the formation of new, nonconserved, repeats. It has also been observed that well-conserved repeats tend to be embedded in protein regions that evolve more slowly than regions containing nonconserved repeats (Hancock et al. 2001; Faux et al. 2007; Mularoni et al. 2007; Simon and Hancock 2009). This bias suggests that repeat evolution is influenced by selection. In order to measure the contribution of natural selection to amino acid tandem repeat evolution, we have compared the level of conservation of human amino acid repeats in orthologous proteins from 12 vertebrate species with the level of conservation of sequences of similar DNA composition but located in noncoding genomic regions, and thus expected to evolve neutrally. We find that the length and degree of phylogenetic conservation of amino acid tandem repeats is much higher than expected under neutrality, and that selection has played an important role in amino acid tandem repeat evolution.