Alexey S. Kondrashov

A method and server for predicting damaging missense mutations

To the Editor: Applications of rapidly advancing sequencing technologies exacerbate the need to interpret individual sequence variants. Sequencing of phenotyped clinical subjects will soon become a method of choice in studies of the genetic causes of Mendelian and complex diseases. New exon capture techniques will direct sequencing efforts towards the most informative and easily interpretable protein-coding fraction of the genome. Thus, the demand for computational predictions of the impact of protein sequence variants will continue to grow. Here we present a new method and the corresponding software tool, PolyPhen-2 (http://genetics.bwh.harvard.edu/pph2/), which is different from the early tool PolyPhen1 in the set of predictive features, alignment pipeline, and the method of classification (Fig. 1a). PolyPhen-2 uses eight sequence-based and three structure-based predictive features (Supplementary Table 1) which were selected automatically by an iterative greedy algorithm (Supplementary Methods). Majority of these features involve comparison of a property of the wild-type (ancestral, normal) allele and the corresponding property of the mutant (derived, disease-causing) allele, which together define an amino acid replacement. Most informative features characterize how well the two human alleles fit into the pattern of amino acid replacements within the multiple sequence alignment of homologous proteins, how distant the protein harboring the first deviation from the human wild-type allele is from the human protein, and whether the mutant allele originated at a hypermutable site2. The alignment pipeline selects the set of homologous sequences for the analysis using a clustering algorithm and then constructs and refines their multiple alignment (Supplementary Fig. 1). The functional significance of an allele replacement is predicted from its individual features (Supplementary Figs. 2–4) by Naive Bayes classifier (Supplementary Methods). Figure 1 PolyPhen-2 pipeline and prediction accuracy. (a) Overview of the algorithm. (b) Receiver operating characteristic (ROC) curves for predictions made by PolyPhen-2 using five-fold cross-validation on HumDiv (red) and HumVar3 (light green). UniRef100 (solid ... We used two pairs of datasets to train and test PolyPhen-2. We compiled the first pair, HumDiv, from all 3,155 damaging alleles with known effects on the molecular function causing human Mendelian diseases, present in the UniProt database, together with 6,321 differences between human proteins and their closely related mammalian homologs, assumed to be non-damaging (Supplementary Methods). The second pair, HumVar3, consists of all the 13,032 human disease-causing mutations from UniProt, together with 8,946 human nsSNPs without annotated involvement in disease, which were treated as non-damaging. We found that PolyPhen-2 performance, as presented by its receiver operating characteristic curves, was consistently superior compared to PolyPhen (Fig. 1b) and it also compared favorably with the three other popular prediction tools4–6 (Fig. 1c). For a false positive rate of 20%, PolyPhen-2 achieves the rate of true positive predictions of 92% and 73% on HumDiv and HumVar, respectively (Supplementary Table 2). One reason for a lower accuracy of predictions on HumVar is that nsSNPs assumed to be non-damaging in HumVar contain a sizable fraction of mildly deleterious alleles. In contrast, most of amino acid replacements assumed non-damaging in HumDiv must be close to selective neutrality. Because alleles that are even mildly but unconditionally deleterious cannot be fixed in the evolving lineage, no method based on comparative sequence analysis is ideal for discriminating between drastically and mildly deleterious mutations, which are assigned to the opposite categories in HumVar. Another reason is that HumDiv uses an extra criterion to avoid possible erroneous annotations of damaging mutations. For a mutation, PolyPhen-2 calculates Naive Bayes posterior probability that this mutation is damaging and reports estimates of false positive (the chance that the mutation is classified as damaging when it is in fact non-damaging) and true positive (the chance that the mutation is classified as damaging when it is indeed damaging) rates. A mutation is also appraised qualitatively, as benign, possibly damaging, or probably damaging (Supplementary Methods). The user can choose between HumDiv- and HumVar-trained PolyPhen-2. Diagnostics of Mendelian diseases requires distinguishing mutations with drastic effects from all the remaining human variation, including abundant mildly deleterious alleles. Thus, HumVar-trained PolyPhen-2 should be used for this task. In contrast, HumDiv-trained PolyPhen-2 should be used for evaluating rare alleles at loci potentially involved in complex phenotypes, dense mapping of regions identified by genome-wide association studies, and analysis of natural selection from sequence data, where even mildly deleterious alleles must be treated as damaging.

Interactions among quantitative traits in the course of sympatric speciation

Sympatric speciation, the origin of two or more species from a single local population, has almost certainly been involved in formation of several species flocks, and may be fairly common in nature. The most straightforward scenario for sympatric speciation requires disruptive selection favouring two substantially different phenotypes, and consists of the evolution of reproductive isolation between them followed by the elimination of all intermediate phenotypes. Here we use the hypergeometric phenotypic model to show that sympatric speciation is possible even when fitness and mate choice depend on different quantitative traits, so that speciation must involve formation of covariance between these traits. The increase in the number of variable lociaffecting fitness facilitates sympatric speciation, whereas the increase in the number of variable loci affecting mate choice has the opposite effect. These predictions may enable more cases of sympatric speciation to be identified.

Selection against harmful mutations in large sexual and asexual populations.

Selection against harmful mutations in large populations is studied assuming that the rate of fitness decrease grows with every new mutation added to a genome. Under this reasonable assumption (Mayr, 1970) the average fitness of a sexual population, without linkage between the loci, is higher, and the average equilibrium number of harmful mutations per individual lower, than in an asexual population. If a gamete contains on the average one or more new mutations, the resulting advantage of sexual reproduction and recombination seems to be sufficient to counterbalance the double advantage of parthenogenesis. Moreover, selection against harmful mutations is probably the most powerful factor preventing linkage disequilibrium even with epistatic interaction between the loci.

The Impact of Comparative Genomics on Our Understanding of Evolution

If an essential gene is displaced by a horizontally transferred one (be it related or unrelated) with the same function, the latter should for the most part be adaptive (confer a distinct evolutionary advantage to the organism) to be fixed in the population. Generally, an alien gene is likely to be at a disadvantage compared to a resident, well-integrated gene with the same function. The reverse may be true, however, after a significant change of the environment, which could apply to the origin of bacterial hyperthermophily via continuous acquisition of archaeal genes. At least one case where the nature of the advantage is clear has been described—eukaryotic isoleucyl-tRNA synthetase confers antibiotic resistance to bacteria whose genome it invades via horizontal transfer (Brown et al. 1998xBrown, J.R, Zhang, J, and Hodgson, J.E. Curr. Biol. 1998; 8: R365–R367Abstract | Full Text | Full Text PDF | PubMedSee all ReferencesBrown et al. 1998). Comparative genome analysis reveals other features that appear to be adaptive, such as lineage-specific expansion of gene families. A striking example of this is the expansion of the photosensory PAS domain in the photosynthetic cyanobacterial lineage, which is without any precedence in nonphotosynthetic organisms. More frequently, however, the teleology behind the presumed adaptation remains unclear.More generally, what we realize with due humility from our first forays into complete-genome-scale comparative genomics, is that we do not truly understand the connection between the genome and the phenotype of an organism. Comparative analysis of the genomes of archaeal and bacterial thermophiles reveals numerous interesting features, including strong evidence of extensive, specific horizontal gene transfer, but fails to identify the genomic basis for thermophily. Similarly, a comparison of the genome of the extreme radioresistant bacterium Deinococcus radiodurans to those of other bacteria showed many peculiarities in its DNA repair and stress response systems (White et al. 1999xWhite, O, Eisen, J.A, Heidelberg, J.F, Hickey, E.K, Peterson, J.D, Dodson, R.J, Haft, D.H, Gwinn, M.L, Nelson, W.C, Richardson, D.L et al. Science. 1999; 286: 1571–1577CrossRef | PubMed | Scopus (627)See all ReferencesWhite et al. 1999), but failed to explain how this organism survives under irradiation of an intensity that makes pyrex glass crack. We believe that the problem of the genome–phenotype connection, which, in a sense, is the central theme of biology, can be solved only through an experimental program strategically planned on the basis of comparative-genomic results. Much of the biological research of the next few decades is likely to develop along these lines.*To whom correspondence should be addressed (e-mail: koonin@ncbi.nlm.nih.gov).

Dobzhansky–Muller incompatibilities in protein evolution

We study fitness landscape in the space of protein sequences by relating sets of human pathogenic missense mutations in 32 proteins to amino acid substitutions that occurred in the course of evolution of these proteins. On average, ≈10% of deviations of a nonhuman protein from its human ortholog are compensated pathogenic deviations (CPDs), i.e., are caused by an amino acid substitution that, at this site, would be pathogenic to humans. Normal functioning of a CPD-containing protein must be caused by other, compensatory deviations of the nonhuman species from humans. Together, a CPD and the corresponding compensatory deviation form a Dobzhansky–Muller incompatibility that can be visualized as the corner on a fitness ridge. Thus, proteins evolve along fitness ridges which contain only ≈10 steps between successive corners. The fraction of CPDs among all deviations of a protein from its human ortholog does not increase with the evolutionary distance between the proteins, indicating that substitutions that carry evolving proteins around these corners occur in rapid succession, driven by positive selection. Data on fitness of interspecies hybrids suggest that the compensatory change that makes a CPD fit usually occurs within the same protein. Data on protein structures and on cooccurrence of amino acids at different sites of multiple orthologous proteins often make it possible to provisionally identify the substitution that compensates a particular CPD.

A universal trend of amino acid gain and loss in protein evolution

Amino acid composition of proteins varies substantially between taxa and, thus, can evolve. For example, proteins from organisms with (G + C)-rich (or (A + T)-rich) genomes contain more (or fewer) amino acids encoded by (G + C)-rich codons. However, no universal trends in ongoing changes of amino acid frequencies have been reported. We compared sets of orthologous proteins encoded by triplets of closely related genomes from 15 taxa representing all three domains of life (Bacteria, Archaea and Eukaryota), and used phylogenies to polarize amino acid substitutions. Cys, Met, His, Ser and Phe accrue in at least 14 taxa, whereas Pro, Ala, Glu and Gly are consistently lost. The same nine amino acids are currently accrued or lost in human proteins, as shown by analysis of non-synonymous single-nucleotide polymorphisms. All amino acids with declining frequencies are thought to be among the first incorporated into the genetic code; conversely, all amino acids with increasing frequencies, except Ser, were probably recruited late. Thus, expansion of initially under-represented amino acids, which began over 3,400 million years ago, apparently continues to this day.

Coevolution of costly mate choice and condition-dependent display of good genes.

Females often choose their mates, instead of mating at random, even when a father contributes nothing but genes to his offspring. Costly female preferences for males with exaggerated traits that reduce viability, such as the peacocks tail, are particularly puzzling. Such preferences can evolve if directly favoured by natural selection or when the exaggerated trait, although maladaptive per se, indicates high overall quality of the males genotype. Two recent analyses suggested that the advantage to mate choice based on genetic quality is too weak to explain extreme cases of exaggeration of display traits and the corresponding preferences. We studied coevolution of a female mate-preference function and a genotype–dependent male display function where mutation supplies variation in genotype quality and mate preference is costly. Preference readily evolves, often causing extreme exaggeration of the display. Mate choice and trait expression can approach an equilibrium, or a limit cycle, or exaggeration can proceed forever, eventually causing extinction.

Genotype-environment interactions and the estimation of the genomic mutation rate in Drosophila melanogaster

We have studied the relative fitnesses of three genotypes of Drosophila melanogaster in 50 environments. Two genotypes, the MA lines, had accumulated mutations in the absence of natural selection over 62 generations. The third was a related strain where selection had continued to act. The environments differed in three factors: parental density, dilution of the medium, and the temperature régime and medium composition. Our measure of fitness assessed fecundity and viability relative to a reference genotype. Both MA lines always had lower fitnesses than the selected line, but the difference increased dramatically with dilution of the medium and, especially, crowding. Under the most severe conditions, the performance of the MA lines approached 0. This increased difference in harsh conditions may be caused both by a uniform increase in the magnitude of deleterious effects of all mutations and by the exposure of mutations which are essentially neutral under benign conditions. If the second cause is important, previous experiments are likely to have underestimated the genomic deleterious mutation rate in Drosophila melanogaster more than previously thought.

Genomic evidence for ameiotic evolution in the bdelloid rotifer Adineta vaga

Loss of sexual reproduction is considered an evolutionary dead end for metazoans, but bdelloid rotifers challenge this view as they appear to have persisted asexually for millions of years. Neither male sex organs nor meiosis have ever been observed in these microscopic animals: oocytes are formed through mitotic divisions, with no reduction of chromosome number and no indication of chromosome pairing. However, current evidence does not exclude that they may engage in sex on rare, cryptic occasions. Here we report the genome of a bdelloid rotifer, Adineta vaga (Davis, 1873), and show that its structure is incompatible with conventional meiosis. At gene scale, the genome of A. vaga is tetraploid and comprises both anciently duplicated segments and less divergent allelic regions. However, in contrast to sexual species, the allelic regions are rearranged and sometimes even found on the same chromosome. Such structure does not allow meiotic pairing; instead, we find abundant evidence of gene conversion, which may limit the accumulation of deleterious mutations in the absence of meiosis. Gene families involved in resistance to oxidation, carbohydrate metabolism and defence against transposons are significantly expanded, which may explain why transposable elements cover only 3% of the assembled sequence. Furthermore, 8% of the genes are likely to be of non-metazoan origin and were probably acquired horizontally. This apparent convergence between bdelloids and prokaryotes sheds new light on the evolutionary significance of sex.

On the origin of species by means of assortative mating.

Assortative mating may split a population even in the absence of natural selection. Here, we study when this happens if mating depends on one or two quantitative traits. Not surprisingly, the modes of assortative mating that can cause sympatric speciation without selection are rather strict. However, some of them may occur in nature. Slow elimination of intermediate individuals caused by the gradual tightening of assortative mating, which evolves owing to relatively weak disruptive selection, provides the alternative scenario for sympatric speciation, in addition to fast elimination of intermediate individuals as a result of the direct action of strong disruptive selection under an invariant mode of assortative mating. Even when assortative mating alone cannot split an initially coherent population, it may be able to prevent the merging of species after their secondary contact.