Identifying protein-coding genes in genomic sequences

The vast majority of the biology of a newly sequenced genome is inferred from the set of encoded proteins. Predicting this set is therefore invariably the first step after the completion of the genome DNA sequence. Here we review the main computational pipelines used to generate the human reference protein-coding gene sets.

Estimation of protein coding density in a corpus of DNA sequence data

A number of experimental methods have been reported for estimating the number of genes in a genome, or the closely related coding density of a genome, defined as the fraction of base pairs in codons. Recently, DNA sequence data representative of the genome as a whole have become available for several organisms, making the problem of estimating coding density amenable to sequence analytic methods. Estimates of coding density for a single genome vary widely, so that methods with characterized error bounds have become increasingly desirable. We present a method to estimate the protein coding density in a corpus of DNA sequence data, in which a ‘coding statistic’ is calculated for a large number of windows of the sequence under study, and the distribution of the statistic is decomposed into two normal distributions, assumed to be the distributions of the coding statistic in the coding and noncoding fractions of the sequence windows. The accuracy of the method is evaluated using known data and application is made to the yeast chromosome III sequence and to C.elegans cosmid sequences. It can also be applied to fragmentary data, for example a collection of short sequences determined in the course of STS mapping.

SGP-1: prediction and validation of homologous genes based on sequence alignments

Conventional methods of gene prediction rely on the recognition of DNA-sequence signals, the coding potential or the comparison of a genomic sequence with a cDNA, EST, or protein database. Reasons for limited accuracy in many circumstances are species-specific training and the incompleteness of reference databases. Lately, comparative genome analysis has attracted increasing attention. Several analysis tools that are based on human/mouse comparisons are already available. Here, we present a program for the prediction of protein-coding genes, termed SGP-1 (Syntenic Gene Prediction), which is based on the similarity of homologous genomic sequences. In contrast to most existing tools, the accuracy of SGP-1 depends little on species-specific properties such as codon usage or the nucleotide distribution. SGP-1 may therefore be applied to nonstandard model organisms in vertebrates as well as in plants, without the need for extensive parameter training. In addition to predicting genes in large-scale genomic sequences, the program may be useful to validate gene structure annotations from databases. To this end, SGP-1 output also contains comparisons between predicted and annotated gene structures in HTML format. The program can be accessed via a Web server at http://soft.ice.mpg.de/sgp-1. The source code, written in ANSI C, is available on request from the authors.

Comparative gene finding in chicken indicates that we are closing in on the set of multi-exonic widely-expressed human genes

The recent availability of the chicken genome sequence poses the question of whether there are human protein-coding genes conserved in chicken that are currently not included in the human gene catalog. Here, we show, using comparative gene finding followed by experimental verification of exon pairs by RT–PCR, that the addition to the multi-exonic subset of this catalog could be as little as 0.2%, suggesting that we may be closing in on the human gene set. Our protocol, however, has two shortcomings: (i) the bioinformatic screening of the predicted genes, applied to filter out false positives, cannot handle intronless genes; and (ii) the experimental verification could fail to identify expression at a specific developmental time. This highlights the importance of developing methods that could provide a reliable estimate of the number of these two types of genes.

Pseudogenes in the ENCODE regions: consensus annotation, analysis of transcription, and evolution

Arising from either retrotransposition or genomic duplication of functional genes, pseudogenes are “genomic fossils” valuable for exploring the dynamics and evolution of genes and genomes. Pseudogene identification is an important problem in computational genomics, and is also critical for obtaining an accurate picture of a genome’s structure and function. However, no consensus computational scheme for defining and detecting pseudogenes has been developed thus far. As part of the ENCyclopedia Of DNA Elements (ENCODE) project, we have compared several distinct pseudogene annotation strategies and found that different approaches and parameters often resulted in rather distinct sets of pseudogenes. We subsequently developed a consensus approach for annotating pseudogenes (derived from protein coding genes) in the ENCODE regions, resulting in 201 pseudogenes, two-thirds of which originated from retrotransposition. A survey of orthologs for these pseudogenes in 28 vertebrate genomes showed that a significant fraction (∼80%) of the processed pseudogenes are primate-specific sequences, highlighting the increasing retrotransposition activity in primates. Analysis of sequence conservation and variation also demonstrated that most pseudogenes evolve neutrally, and processed pseudogenes appear to have lost their coding potential immediately or soon after their emergence. In order to explore the functional implication of pseudogene prevalence, we have extensively examined the transcriptional activity of the ENCODE pseudogenes. We performed systematic series of pseudogene-specific RACE analyses. These, together with complementary evidence derived from tiling microarrays and high throughput sequencing, demonstrated that at least a fifth of the 201 pseudogenes are transcribed in one or more cell lines or tissues.

Improving gene annotation using peptide mass spectrometry

Annotation of protein-coding genes is a key goal of genome sequencing projects. In spite of tremendous recent advances in computational gene finding, comprehensive annotation remains a challenge. Peptide mass spectrometry is a powerful tool for researching the dynamic proteome and suggests an attractive approach to discover and validate protein-coding genes. We present algorithms to construct and efficiently search spectra against a genomic database, with no prior knowledge of encoded proteins. By searching a corpus of 18.5 million tandem mass spectra (MS/MS) from human proteomic samples, we validate 39,000 exons and 11,000 introns at the level of translation. We present translation-level evidence for novel or extended exons in 16 genes, confirm translation of 224 hypothetical proteins, and discover or confirm over 40 alternative splicing events. Polymorphisms are efficiently encoded in our database, allowing us to observe variant alleles for 308 coding SNPs. Finally, we demonstrate the use of mass spectrometry to improve automated gene prediction, adding 800 correct exons to our predictions using a simple rescoring strategy. Our results demonstrate that proteomic profiling should play a role in any genome sequencing project.

Prominent use of distal 5’ transcription start sites and discovery of a large number of additional exons in ENCODE regions

This report presents systematic empirical annotation of transcript products from 399 annotated protein-coding loci across the 1% of the human genome targeted by the Encyclopedia of DNA elements (ENCODE) pilot project using a combination of 5' rapid amplification of cDNA ends (RACE) and high-density resolution tiling arrays. We identified previously unannotated and often tissue- or cell-line-specific transcribed fragments (RACEfrags), both 5' distal to the annotated 5' terminus and internal to the annotated gene bounds for the vast majority (81.5%) of the tested genes. Half of the distal RACEfrags span large segments of genomic sequences away from the main portion of the coding transcript and often overlap with the upstream-annotated gene(s). Notably, at least 20% of the resultant novel transcripts have changes in their open reading frames (ORFs), most of them fusing ORFs of adjacent transcripts. A significant fraction of distal RACEfrags show expression levels comparable to those of known exons of the same locus, suggesting that they are not part of very minority splice forms. These results have significant implications concerning (1) our current understanding of the architecture of protein-coding genes; (2) our views on locations of regulatory regions in the genome; and (3) the interpretation of sequence polymorphisms mapping to regions hitherto considered to be "noncoding," ultimately relating to the identification of disease-related sequence alterations.

GENCODE: producing a reference annotation for ENCODE

Background: The GENCODE consortium was formed to identify and map all protein-coding genes within the ENCODE regions. This was achieved by a combination of initial manualannotation by the HAVANA team, experimental validation by the GENCODE consortium and a refinement of the annotation based on these experimental results.Results: The GENCODE gene features are divided into eight different categories of which onlythe first two (known and novel coding sequence) are confidently predicted to be protein-codinggenes. 5’ rapid amplification of cDNA ends (RACE) and RT-PCR were used to experimentallyverify the initial annotation. Of the 420 coding loci tested, 229 RACE products have beensequenced. They supported 5’ extensions of 30 loci and new splice variants in 50 loci. In addition,46 loci without evidence for a coding sequence were validated, consisting of 31 novel and 15putative transcripts. We assessed the comprehensiveness of the GENCODE annotation byattempting to validate all the predicted exon boundaries outside the GENCODE annotation. Outof 1,215 tested in a subset of the ENCODE regions, 14 novel exon pairs were validated, only twoof them in intergenic regions.Conclusions: In total, 487 loci, of which 434 are coding, have been annotated as part of theGENCODE reference set available from the UCSC browser. Comparison of GENCODEannotation with RefSeq and ENSEMBL show only 40% of GENCODE exons are contained withinthe two sets, which is a reflection of the high number of alternative splice forms with uniqueexons annotated. Over 50% of coding loci have been experimentally verified by 5’ RACE forEGASP and the GENCODE collaboration is continuing to refine its annotation of 1% humangenome with the aid of experimental validation.

EGASP: the human ENCODE Genome Annotation Assessment Project

Background: We present the results of EGASP, a community experiment to assess the state-ofthe-art in genome annotation within the ENCODE regions, which span 1% of the human genomesequence. The experiment had two major goals: the assessment of the accuracy of computationalmethods to predict protein coding genes; and the overall assessment of the completeness of thecurrent human genome annotations as represented in the ENCODE regions. For thecomputational prediction assessment, eighteen groups contributed gene predictions. Weevaluated these submissions against each other based on a ‘reference set’ of annotationsgenerated as part of the GENCODE project. These annotations were not available to theprediction groups prior to the submission deadline, so that their predictions were blind and anexternal advisory committee could perform a fair assessment.Results: The best methods had at least one gene transcript correctly predicted for close to 70%of the annotated genes. Nevertheless, the multiple transcript accuracy, taking into accountalternative splicing, reached only approximately 40% to 50% accuracy. At the coding nucleotidelevel, the best programs reached an accuracy of 90% in both sensitivity and specificity. Programsrelying on mRNA and protein sequences were the most accurate in reproducing the manuallycurated annotations. Experimental validation shows that only a very small percentage (3.2%) of the selected 221 computationally predicted exons outside of the existing annotation could beverified.Conclusions: This is the first such experiment in human DNA, and we have followed thestandards established in a similar experiment, GASP1, in Drosophila melanogaster. We believe theresults presented here contribute to the value of ongoing large-scale annotation projects and shouldguide further experimental methods when being scaled up to the entire human genome sequence.

Design and evaluation of a panel os single-nucleotide polymorphisms in microRNA genomic regions for association studies in human disease

MicroRNAs (miRNA) are recognized posttranscriptional gene repressors involved in the control of almost every biological process. Allelic variants in these regions may be an important source of phenotypic diversity and contribute to disease susceptibility. We analyzed the genomic organization of 325 human miRNAs (release 7.1, miRBase) to construct a panel of 768 single-nucleotide polymorphisms (SNPs) covering approximately 1 Mb of genomic DNA, including 131 isolated miRNAs (40%) and 194 miRNAs arranged in 48 miRNA clusters, as well as their 5-kb flanking regions. Of these miRNAs, 37% were inside known protein-coding genes, which were significantly associated with biological functions regarding neurological, psychological or nutritional disorders. SNP coverage analysis revealed a lower SNP density in miRNAs compared with the average of the genome, with only 24 SNPs located in the 325 miRNAs studied. Further genotyping of 340 unrelated Spanish individuals showed that more than half of the SNPs in miRNAs were either rare or monomorphic, in agreement with the reported selective constraint on human miRNAs. A comparison of the minor allele frequencies between Spanish and HapMap population samples confirmed the applicability of this SNP panel to the study of complex disorders among the Spanish population, and revealed two miRNA regions, hsa-mir-26a-2 in the CTDSP2 gene and hsa-mir-128-1 in the R3HDM1 gene, showing geographical allelic frequency variation among the four HapMap populations, probably because of differences in natural selection. The designed miRNA SNP panel could help to identify still hidden links between miRNAs and human disease.

Do protein–protein interaction databases identify moonlighting proteins?

One of the most striking results of the human (and mammalian) genomes is the low number of protein-coding genes. To-date, the main molecular mechanism to increase the number of different protein isoforms and functions is alternative splicing. However, a less-known way to increase the number of protein functions is the existence of multifunctional, multitask, or ‘‘moonlighting’’, proteins. By and large, moonlighting proteins are experimentally disclosed by serendipity. Proteomics is becoming one of the very active areas of biomedical research, which permits researchers to identify previously unseen connections among proteins and pathways. In principle, protein–protein interaction (PPI) databases should contain information on moonlighting proteins and could provide suggestions to further analysis in order to prove the multifunctionality. As far as we know, nobody has verified whether PPI databases actually disclose moonlighting proteins. In the present work we check whether well-established moonlighting proteins present in PPI databases connect with their known partners and, therefore, a careful inspection of these databases could help to suggest their different functions. The results of our research suggest that PPI databases could be a valuable tool to suggest multifunctionality.

The genome and development-dependent transcriptomes of Pyronema confluens: a window into fungal evolution

Fungi are a large group of eukaryotes found in nearly all ecosystems. More than 250 fungal genomes have already been sequenced, greatly improving our understanding of fungal evolution, physiology, and development. However, for the Pezizomycetes, an early-diverging lineage of filamentous ascomycetes, there is so far only one genome available, namely that of the black truffle, Tuber melanosporum, a mycorrhizal species with unusual subterranean fruiting bodies. To help close the sequence gap among basal filamentous ascomycetes, and to allow conclusions about the evolution of fungal development, we sequenced the genome and assayed transcriptomes during development of Pyronema confluens, a saprobic Pezizomycete with a typical apothecium as fruiting body. With a size of 50 Mb and ~13,400 protein-coding genes, the genome is more characteristic of higher filamentous ascomycetes than the large, repeat-rich truffle genome; however, some typical features are different in the P. confluens lineage, e.g. the genomic environment of the mating type genes that is conserved in higher filamentous ascomycetes, but only partly conserved in P. confluens. On the other hand, P. confluens has a full complement of fungal photoreceptors, and expression studies indicate that light perception might be similar to distantly related ascomycetes and, thus, represent a basic feature of filamentous ascomycetes. Analysis of spliced RNA-seq sequence reads allowed the detection of natural antisense transcripts for 281 genes. The P. confluens genome contains an unusually high number of predicted orphan genes, many of which are upregulated during sexual development, consistent with the idea of rapid evolution of sex-associated genes. Comparative transcriptomics identified the transcription factor gene pro44 that is upregulated during development in P. confluens and the Sordariomycete Sordaria macrospora. The P. confluens pro44 gene (PCON_06721) was used to complement the S. macrospora pro44 deletion mutant, showing functional conservation of this developmental regulator.

Complete DNA sequence of Kuraishia capsulata illustrates novel genomic features among budding yeasts (Saccharomycotina)

The numerous yeast genome sequences presently available provide a rich source of information for functional as well as evolutionary genomics but unequally cover the large phylogenetic diversity of extant yeasts. We present here the complete sequence of the nuclear genome of the haploid-type strain of Kuraishia capsulata (CBS1993(T)), a nitrate-assimilating Saccharomycetales of uncertain taxonomy, isolated from tunnels of insect larvae underneath coniferous barks and characterized by its copious production of extracellular polysaccharides. The sequence is composed of seven scaffolds, one per chromosome, totaling 11.4 Mb and containing 6,029 protein-coding genes, ~13.5% of which being interrupted by introns. This GC-rich yeast genome (45.7%) appears phylogenetically related with the few other nitrate-assimilating yeasts sequenced so far, Ogataea polymorpha, O. parapolymorpha, and Dekkera bruxellensis, with which it shares a very reduced number of tRNA genes, a novel tRNA sparing strategy, and a common nitrate assimilation cluster, three specific features to this group of yeasts. Centromeres were recognized in GC-poor troughs of each scaffold. The strain bears MAT alpha genes at a single MAT locus and presents a significant degree of conservation with Saccharomyces cerevisiae genes, suggesting that it can perform sexual cycles in nature, although genes involved in meiosis were not all recognized. The complete absence of conservation of synteny between K. capsulata and any other yeast genome described so far, including the three other nitrate-assimilating species, validates the interest of this species for long-range evolutionary genomic studies among Saccharomycotina yeasts.

The genome of the recently domesticated crop plant sugar beet (Beta vulgaris)

Sugar beet (Beta vulgaris ssp. vulgaris) is an important crop of temperate climates which provides nearly 30% of the world's annual sugar production and is a source for bioethanol and animal feed. The species belongs to the order of Caryophylalles, is diploid with 2n = 18 chromosomes, has an estimated genome size of 714-758 megabases and shares an ancient genome triplication with other eudicot plants. Leafy beets have been cultivated since Roman times, but sugar beet is one of the most recently domesticated crops. It arose in the late eighteenth century when lines accumulating sugar in the storage root were selected from crosses made with chard and fodder beet. Here we present a reference genome sequence for sugar beet as the first non-rosid, non-asterid eudicot genome, advancing comparative genomics and phylogenetic reconstructions. The genome sequence comprises 567 megabases, of which 85% could be assigned to chromosomes. The assembly covers a large proportion of the repetitive sequence content that was estimated to be 63%. We predicted 27,421 protein-coding genes supported by transcript data and annotated them on the basis of sequence homology. Phylogenetic analyses provided evidence for the separation of Caryophyllales before the split of asterids and rosids, and revealed lineage-specific gene family expansions and losses. We sequenced spinach (Spinacia oleracea), another Caryophyllales species, and validated features that separate this clade from rosids and asterids. Intraspecific genomic variation was analysed based on the genome sequences of sea beet (Beta vulgaris ssp. maritima; progenitor of all beet crops) and four additional sugar beet accessions. We identified seven million variant positions in the reference genome, and also large regions of low variability, indicating artificial selection. The sugar beet genome sequence enables the identification of genes affecting agronomically relevant traits, supports molecular breeding and maximizes the plant's potential in energy biotechnology.

Acoel flatworms are not Platyhelminthes: evidence from phylogenomics

Acoel flatworms are small marine worms traditionally considered to belong to the phylum Platyhelminthes. However, molecular phylogenetic analyses suggest that acoels are not members of Platyhelminthes, but are rather extant members of the earliest diverging Bilateria. This result has been called into question, under suspicions of a long branch attraction (LBA) artefact. Here we re-examine this problem through a phylogenomic approach using 68 different protein-coding genes from the acoel Convoluta pulchra and 51 metazoan species belonging to 15 different phyla. We employ a mixture model, named CAT, previously found to overcome LBA artefacts where classical models fail. Our results unequivocally show that acoels are not part of the classically defined Platyhelminthes, making the latter polyphyletic. Moreover, they indicate a deuterostome affinity for acoels, potentially as a sister group to all deuterostomes, to Xenoturbellida, to Ambulacraria, or even to chordates. However, the weak support found for most deuterostome nodes, together with the very fast evolutionary rate of the acoel Convoluta pulchra, call for more data from slowly evolving acoels (or from its sister-group, the Nemertodermatida) to solve this challenging phylogenetic problem.