Length of the ORF, position of the first AUG and the Kozak motif are important factors in potential dual-coding transcripts

Class similarity network for coding and long non-coding RNA classification

BACKGROUND

Long non-coding RNAs (lncRNAs) play significant roles in varieties of physiological and pathological processes.The premise of the lncRNA functional study is that the lncRNAs are identified correctly. Recently, deep learning method like convolutional neural network (CNN) has been successfully applied to identify the lncRNAs. However, the traditional CNN considers little relationships among samples via an indirect way.

RESULTS

Inspired by the Siamese Neural Network (SNN), here we propose a novel network named Class Similarity Network in coding RNA and lncRNA classification. Class Similarity Network considers more relationships among input samples in a direct way. It focuses on exploring the potential relationships between input samples and samples from both the same class and the different classes. To achieve this, Class Similarity Network trains the parameters specific to each class to obtain the high-level features and represents the general similarity to each class in a node. The comparison results on the validation dataset under the same conditions illustrate the superiority of our Class Similarity Network to the baseline CNN. Besides, our method performs effectively and achieves state-of-the-art performances on two test datasets.

CONCLUSIONS

We construct Class Similarity Network in coding RNA and lncRNA classification, which is shown to work effectively on two different datasets by achieving accuracy, precision, and F1-score as 98.43%, 0.9247, 0.9374, and 97.54%, 0.9990, 0.9860, respectively.

Common Features in lncRNA Annotation and Classification: A Survey

Long non-coding RNAs (lncRNAs) are widely recognized as important regulators of gene expression. Their molecular functions range from miRNA sponging to chromatin-associated mechanisms, leading to effects in disease progression and establishing them as diagnostic and therapeutic targets. Still, only a few representatives of this diverse class of RNAs are well studied, while the vast majority is poorly described beyond the existence of their transcripts. In this review we survey common in silico approaches for lncRNA annotation. We focus on the well-established sets of features used for classification and discuss their specific advantages and weaknesses. While the available tools perform very well for the task of distinguishing coding sequence from other RNAs, we find that current methods are not well suited to distinguish lncRNAs or parts thereof from other non-protein-coding input sequences. We conclude that the distinction of lncRNAs from intronic sequences and untranslated regions of coding mRNAs remains a pressing research gap.

FexRNA: Exploratory Data Analysis and Feature Selection of Non-Coding RNA

Non-coding RNA (ncRNA) is involved in many biological processes and diseases in all species. Many ncRNA datasets exist that provide ncRNA data in FASTA format which is well suited for biomedical purposes. However, for ncRNA analysis and classification, statistical learning methods require hidden numerical features from the data. Furthermore, in the literature, a wealth of sequence intrinsic features has been proposed for ncRNA identification. The extraction of hidden features, their analysis, and usage of a suitable set of features is crucial for the performance of any statistical learning method. To alleviate the posed challenges, we generated 96 feature datasets from ncRNA widely used features. The feature datasets are based on RNACentral and consist of species, ncRNA types, and expert databases that are available on the FexRNA platform. Additionally, the feature datasets are explored and analysed to provide statistical information, univariate, and bivariate analysis. We sought to determine which of these 17 features would be most appropriate to use in developing ncRNA classification approaches. For feature selection (FS), a two-phase hierarchical FS framework based on correlation and majority voting is proposed and evaluated on 5 species. The FexRNA platform provides information about ncRNA feature analysis and selection.

Understanding small ORF diversity through a comprehensive transcription feature classification

Small open reading frames (small ORFs/sORFs/smORFs) are potentially coding sequences smaller than 100 codons that have historically been considered junk DNA by gene prediction software and in annotation screening; however, the advent of next-generation sequencing has contributed to the deeper investigation of junk DNA regions and their transcription products, resulting in the emergence of smORFs as a new focus of interest in systems biology. Several smORF peptides were recently reported in noncanonical mRNAs as new players in numerous biological contexts; however, their relevance is still overlooked in coding potential analysis. Hence, this review proposes a smORF classification based on transcriptional features, discussing the most promising approaches to investigate smORFs based on their different characteristics. First, smORFs were divided into nonexpressed (intergenic) and expressed (genic) smORFs. Second, genic smORFs were classified as smORFs located in noncoding RNAs (ncRNAs) or canonical mRNAs. Finally, smORFs in ncRNAs were further subdivided into sequences located in small or long RNAs, whereas smORFs located in canonical mRNAs were subdivided into several specific classes depending on their localization along the gene. We hope that this review provides new insights into large-scale annotations and reinforces the role of smORFs as essential components of a hidden coding DNA world.

New insights into the evolutionary features of viral overlapping genes by discriminant analysis

Overlapping genes originate by a mechanism of overprinting, in which nucleotide substitutions in a pre-existing frame induce the expression of a de novo protein from an alternative frame. In this study, I assembled a dataset of 319 viral overlapping genes, which included 82 overlaps whose expression is experimentally known and the respective 237 homologs. Principal component analysis revealed that overlapping genes have a common pattern of nucleotide and amino acid composition. Discriminant analysis separated overlapping from non-overlapping genes with an accuracy of 97%. When applied to overlapping genes with known genealogy, it separated ancestral from de novo frames with an accuracy close to 100%. This high discriminant power was crucial to computationally design variants of de novo viral proteins known to possess selective anticancer toxicity (apoptin) or protection against neurodegeneration (X protein), as well as to detect two new potential overlapping genes in the genome of the new coronavirus SARS-CoV-2.

Detection and Verification of Mammalian Mirtrons by Northern Blotting

microRNAs (miRNAs) have vital roles in regulating gene expression-contributing to major diseases like cancer and heart disease. Over the last decade, thousands of miRNAs have been discovered through high throughput sequencing-based annotation. Different classes have been described, as well as a great dynamic range of expression levels. While sequencing approaches provide insight into biogenesis and allow confident identification, there is a need for additional methods for validation and characterization. Northern blotting was one of the first techniques used for studying miRNAs, and remains one of the most valuable as it avoids enzymatic manipulation of miRNA transcripts. Blotting can also provide insight into biogenesis by revealing RNA processing intermediates. Compared to sequencing, however, northern blotting is a relatively insensitive technology. This creates a challenge for detecting low expressed miRNAs, particularly those produced by inefficient, non-canonical pathways. In this chapter, we describe a strategy to study such miRNAs by northern blotting that involves ectopic expression of both miRNAs and miRNA-binding Argonaute (Ago) proteins. Through use of epitope tags, this strategy also provides a convenient method for verification of small RNA competency to be loaded into regulatory complexes.

Genome-wide identification and differential analysis of translational initiation

Translation is principally regulated at the initiation stage. The development of the translation initiation (TI) sequencing (TI-seq) technique has enabled the global mapping of TIs and revealed unanticipated complex translational landscapes in metazoans. Despite the wide adoption of TI-seq, there is no computational tool currently available for analyzing TI-seq data. To fill this gap, we develop a comprehensive toolkit named Ribo-TISH, which allows for detecting and quantitatively comparing TIs across conditions from TI-seq data. Ribo-TISH can also predict novel open reading frames (ORFs) from regular ribosome profiling (rRibo-seq) data and outperform several established methods in both computational efficiency and prediction accuracy. Applied to published TI-seq/rRibo-seq data sets, Ribo-TISH uncovers a novel signature of elevated mitochondrial translation during amino-acid deprivation and predicts novel ORFs in 5′UTRs, long noncoding RNAs, and introns. These successful applications demonstrate the power of Ribo-TISH in extracting biological insights from TI-seq/rRibo-seq data.

Translation initiation sequencing (TI-seq) has revealed unexpected diversity in protein isoforms. Here, Zhang et al. present Ribo-TISH, a computational toolkit that can detect and compare TIs across conditions and improve open reading frame prediction from different types of ribosome profiling data.

ExUTR: a novel pipeline for large-scale prediction of 3'-UTR sequences from NGS data

Background

The three prime untranslated region (3′-UTR) is known to play a pivotal role in modulating gene expression by determining the fate of mRNA. Many crucial developmental events, such as mammalian spermatogenesis, tissue patterning, sex determination and neurogenesis, rely heavily on post-transcriptional regulation by the 3′-UTR. However, 3′-UTR biology seems to be a relatively untapped field, with only limited tools and 3′-UTR resources available. To elucidate the regulatory mechanisms of the 3′-UTR on gene expression, firstly the 3′-UTR sequences must be identified. Current 3′-UTR mining tools, such as GETUTR, 3USS and UTRscan, all depend on a well-annotated reference genome or curated 3′-UTR sequences, which hinders their application on a myriad of non-model organisms where the genomes are not available. To address these issues, the establishment of an NGS-based, automated pipeline is urgently needed for genome-wide 3′-UTR prediction in the absence of reference genomes.

Results

Here, we propose ExUTR, a novel NGS-based pipeline to predict and retrieve 3′-UTR sequences from RNA-Seq experiments, particularly designed for non-model species lacking well-annotated genomes. This pipeline integrates cutting-edge bioinformatics tools, databases (Uniprot and UTRdb) and novel in-house Perl scripts, implementing a fully automated workflow. By taking transcriptome assemblies as inputs, this pipeline identifies 3′-UTR signals based primarily on the intrinsic features of transcripts, and outputs predicted 3′-UTR candidates together with associated annotations. In addition, ExUTR only requires minimal computational resources, which facilitates its implementation on a standard desktop computer with reasonable runtime, making it affordable to use for most laboratories. We also demonstrate the functionality and extensibility of this pipeline using publically available RNA-Seq data from both model and non-model species, and further validate the accuracy of predicted 3′-UTR using both well-characterized 3′-UTR resources and 3P–Seq data.

Conclusions

ExUTR is a practical and powerful workflow that enables rapid genome-wide 3′-UTR discovery from NGS data. The candidates predicted through this pipeline will further advance the study of miRNA target prediction, cis elements in 3′-UTR and the evolution and biology of 3′-UTRs. Being independent of a well-annotated reference genome will dramatically expand its application to much broader research area, encompassing all species for which RNA-Seq is available.

Electronic supplementary material

The online version of this article (10.1186/s12864-017-4241-1) contains supplementary material, which is available to authorized users.

Long Noncoding RNA Identification: Comparing Machine Learning Based Tools for Long Noncoding Transcripts Discrimination

Long noncoding RNA (lncRNA) is a kind of noncoding RNA with length more than 200 nucleotides, which aroused interest of people in recent years. Lots of studies have confirmed that human genome contains many thousands of lncRNAs which exert great influence over some critical regulators of cellular process. With the advent of high-throughput sequencing technologies, a great quantity of sequences is waiting for exploitation. Thus, many programs are developed to distinguish differences between coding and long noncoding transcripts. Different programs are generally designed to be utilised under different circumstances and it is sensible and practical to select an appropriate method according to a certain situation. In this review, several popular methods and their advantages, disadvantages, and application scopes are summarised to assist people in employing a suitable method and obtaining a more reliable result.

Thousands of novel translated open reading frames in humans inferred by ribosome footprint profiling

Accurate annotation of protein coding regions is essential for understanding how genetic information is translated into function. We describe riboHMM, a new method that uses ribosome footprint data to accurately infer translated sequences. Applying riboHMM to human lymphoblastoid cell lines, we identified 7273 novel coding sequences, including 2442 translated upstream open reading frames. We observed an enrichment of footprints at inferred initiation sites after drug-induced arrest of translation initiation, validating many of the novel coding sequences. The novel proteins exhibit significant selective constraint in the inferred reading frames, suggesting that many are functional. Moreover, ~40% of bicistronic transcripts showed negative correlation in the translation levels of their two coding sequences, suggesting a potential regulatory role for these novel regions. Despite known limitations of mass spectrometry to detect protein expressed at low level, we estimated a 14% validation rate. Our work significantly expands the set of known coding regions in humans.

DOI:http://dx.doi.org/10.7554/eLife.13328.001

Death of a dogma: eukaryotic mRNAs can code for more than one protein

mRNAs carry the genetic information that is translated by ribosomes. The traditional view of a mature eukaryotic mRNA is a molecule with three main regions, the 5' UTR, the protein coding open reading frame (ORF) or coding sequence (CDS), and the 3' UTR. This concept assumes that ribosomes translate one ORF only, generally the longest one, and produce one protein. As a result, in the early days of genomics and bioinformatics, one CDS was associated with each protein-coding gene. This fundamental concept of a single CDS is being challenged by increasing experimental evidence indicating that annotated proteins are not the only proteins translated from mRNAs. In particular, mass spectrometry (MS)-based proteomics and ribosome profiling have detected productive translation of alternative open reading frames. In several cases, the alternative and annotated proteins interact. Thus, the expression of two or more proteins translated from the same mRNA may offer a mechanism to ensure the co-expression of proteins which have functional interactions. Translational mechanisms already described in eukaryotic cells indicate that the cellular machinery is able to translate different CDSs from a single viral or cellular mRNA. In addition to summarizing data showing that the protein coding potential of eukaryotic mRNAs has been underestimated, this review aims to challenge the single translated CDS dogma.

LncRNA-ID: Long non-coding RNA IDentification using balanced random forests

MOTIVATION

Long non-coding RNAs (lncRNAs), which are non-coding RNAs of length above 200 nucleotides, play important biological functions such as gene expression regulation. To fully reveal the functions of lncRNAs, a fundamental step is to annotate them in various species. However, as lncRNAs tend to encode one or multiple open reading frames, it is not trivial to distinguish these long non-coding transcripts from protein-coding genes in transcriptomic data.

RESULTS

In this work, we design a new tool that calculates the coding potential of a transcript using a machine learning model (random forest) based on multiple features including sequence characteristics of putative open reading frames, translation scores based on ribosomal coverage, and conservation against characterized protein families. The experimental results show that our tool competes favorably with existing coding potential computation tools in lncRNA identification.

AVAILABILITY AND IMPLEMENTATION

The scripts and data can be downloaded at https://github.com/zhangy72/LncRNA-ID.

Found in translation: functions and evolution of a recently discovered alternative proteome

A major goal in biology is to map entire proteomes to better understand the biology and evolution of cells. However, our current views of proteomes are conservative and biased against small proteins. Besides serendipitous discoveries of small proteins, it has been largely assumed that eukaryotic mature mRNAs contain a single ORF and that non-coding RNAs are not translated because their ORFs are too short to play a functional role. A flurry of recent studies brought to light an unexplored proteome that is mainly translated from short ORFs in non-coding regions and from alternative ORFs (AltORFs) in reference genes. The detection of these small proteins and the elucidation of their functions remain challenging and open a new dimension of eukaryotic proteomes, including the birth of novel genes and proteins.

Torularhodin, isolated from Sporidiobolus pararoseus, inhibits human prostate cancer LNCaP and PC-3 cell growth through Bcl-2/Bax mediated apoptosis and AR down-regulation

Torularhodin is one of the principal carotenoids inSporidiobolus pararoseusand has a similar structure to that of lycopene.

Computational analysis and mapping of novel open reading frames in influenza A viruses

The influenza A virus contains 8 segmented genomic RNAs and was considered to encode 10 viral proteins until investigators identified the 11^th viral protein, PB1-F2, which uses an alternative reading frame of the PB1 gene. The recently identified PB1-N40, PA-N155 and PA-N182 influenza A proteins have shown the potential for using a leaking ribosomal scanning mechanism to generate novel open reading frames (ORFs). These novel ORFs provide examples of the manner in which the influenza A virus expands its coding capacity by using overlapping reading frames. In this study, we performed a computational search, based on a ribosome scanning mechanism, on all influenza A coding sequences to identify possible forward-reading ORFs that could be translated into novel viral proteins. We specified that the translated products had a prevalence ≥5% to eliminate sporadic ORFs. A total of 1,982 ORFs were thus identified and presented in terms of their locations, lengths and Kozak sequence strengths. We further provided an abridged list of ORFs by requiring every candidate an upstream start codon (within the upstream third of the primary transcript), a strong Kozak consensus sequence and high prevalence (≥95% and ≥50% for in-frame and alternative-frame ORFs, respectively). The PB1-F2, PB1-N40, PA-N155 and PA-N182 proteins all fulfilled our filtering criteria. Subject to these three stringent settings, we additionally named 16 novel ORFs for all influenza A genomes except for HA and NA, for which 43 HA and 11 NA ORFs from their respective subtypes were also recognized.

The effect of leader peptide mutations on the biological function of bovine myostatin gene

The growth of muscle fibers can be negatively regulated by bovine myostatin. The first two exons of myostatin gene code for the N-propeptide and its third exon codes for the C-polypeptide. Myostatin is secreted as a latent configuration formed by dimerization of two matured C peptides non-covalently linked with the N terminal pro-peptide. Pro-peptide has two distinct functions in guiding protein folding and regulating biological activity of myostatin. When the structure of the leader peptide is altered via mutations resulting in more tight binding with the mature peptide, myostatin function is inhibited, resulting in the changes of P21 and CDK2 expression levels which are related to the regulation of cell cycle. In the present study, the coding region of bMSTN (bovine myostatin) gene was amplified and mutated (A224C and G938A) through fusion PCR, and the mutated bMSTN gene (bMSTN-mut) was inserted in frame into the pEF1a-IRES-DsRed-Express2 vector and transfected into bovine fibroblast cells. The expression levels of bMSTN-mut, P21 and CDK2 (cyclin dependent kinase 2) were examined with qPCR and Western-blotting. Changes in cell cycle after transfection were also analyzed with flow cytometry. The results indicated that leader peptide mutation resulted in down-regulation of P21 expression levels and up-regulation of CDK2 expression levels. The flow cytometry results showed that the proportion of cells in the G0/G1-phase was lower and that of cells in the S-phase was higher in bMSTN-mut transfected group than that in the control group. The proliferation rate of bMSTN-mut transfected cells was also significantly higher than that of the control cells. In conclusion, the studies have shown that the pEF1a-IRES-DsRed-Express2-bMSTN-mut recombinant plasmid could effectively promote the proliferation of bovine fibroblast cells. The site-directed mutagenesis of bMSTN gene leader peptide and in vitro expression in bovine fibroblast cells could be helpful to further the studies of bMSTN in regulating bovine muscle cell growth and development.

Emerging evidence for functional peptides encoded by short open reading frames

Short open reading frames (sORFs) are a common feature of all genomes, but their coding potential has mostly been disregarded, partly because of the difficulty in determining whether these sequences are translated. Recent innovations in computing, proteomics and high-throughput analyses of translation start sites have begun to address this challenge and have identified hundreds of putative coding sORFs. The translation of some of these has been confirmed, although the contribution of their peptide products to cellular functions remains largely unknown. This Review examines this hitherto overlooked component of the proteome and considers potential roles for sORF-encoded peptides.

The A2A adenosine receptor is a dual coding gene: a novel mechanism of gene usage and signal transduction

The A2A adenosine receptor (A2AR) is a G protein-coupled receptor and a major target of caffeine. The A2AR gene encodes alternative transcripts that are initiated from at least two independent promoters. The different transcripts of the A2AR gene contain the same coding region and 3'-untranslated region and different 5'-untranslated regions that are highly conserved among species. We report here that in addition to the production of the A2AR protein, translation from an upstream, out-of-frame AUG of the rat A2AR gene produces a 134-amino acid protein (designated uORF5). An anti-uORF5 antibody recognized a protein of the predicted size of uORF5 in PC12 cells and rat brains. Up-regulation of A2AR transcripts by hypoxia led to increased levels of both the A2AR and uORF5 proteins. Moreover, stimulation of A2AR increased the level of the uORF5 protein via post-transcriptional regulation. Expression of the uORF5 protein suppressed the AP1-mediated transcription promoted by nerve growth factor and modulated the expression of several proteins that were implicated in the MAPK pathway. Taken together, our results show that the rat A2AR gene encodes two distinct proteins (A2AR and uORF5) in an A2AR-dependent manner. Our study reveals a new example of the complexity of the mammalian genome and provides novel insights into the function of A2AR.

A novel splicing outcome reveals more than 2000 new mammalian protein isoforms

MOTIVATION

We have recently characterized an instance of alternative splicing that differs from the canonical gene transcript by deletion of a length of sequence not divisible by three, but where translation can be rescued by an alternative start codon. This results in a predicted protein in which the amino terminus differs markedly in sequence from the known protein product(s), as it is translated from an alternative reading frame. Automated pipelines have annotated thousands of splice variants but have overlooked these protein isoforms, leading to them being underrepresented in current databases.

RESULTS

Here we describe 1849 human and 733 mouse transcripts that can be transcribed from an alternate ATG. Of these, >80% have not been annotated previously. Those conserved between human and mouse genomes (and hence under likely evolutionary selection) are identified. We provide mass spectroscopy evidence for translation of selected transcripts. Of the described splice variants, only one has previously been studied in detail and converted the encoded protein from an activator of cell-function to a suppressor, demonstrating that these splice variants can result in profound functional change. We investigate the potential functional effects of this splicing using a variety of bioinformatic tools. The 2582 variants we describe are involved in a wide variety of biological processes, and therefore open many new avenues of research.

HAltORF: a database of predicted out-of-frame alternative open reading frames in human

Human alternative open reading frames (HAltORF) is a publicly available and searchable online database referencing putative products of out-of-frame alternative translation initiation (ATI) in human mRNAs. Out-of-frame ATI is a process by which a single mRNA encodes independent proteins, when distinct initiation codons located in different reading frames are recognized by a ribosome to initiate translation. This mechanism is largely used in viruses to increase the coding potential of small viral genomes. There is increasing evidence that out-of-frame ATI is also used in eukaryotes, including human, and may contribute to the diversity of the human proteome. HAltORF is the first web-based searchable database that allows thorough investigation in the human transcriptome of out-of-frame alternative open reading frames with a start codon located in a strong Kozak context, and are thus the more likely to be expressed. It is also the first large scale study on the human transcriptome to successfully predict the expression of out-of-frame ATI protein products that were previously discovered experimentally. HAltORF will be a useful tool for the identification of human genes with multiple coding sequences, and will help to better define and understand the complexity of the human proteome.

Database URL:http://haltorf.roucoulab.com/.

Is there a twelfth protein-coding gene in the genome of influenza A? A selection-based approach to the detection of overlapping genes in closely related sequences

Protein-coding genes often contain long overlapping open-reading frames (ORFs), which may or may not be functional. Current methods that utilize the signature of purifying selection to detect functional overlapping genes are limited to the analysis of sequences from divergent species, thus rendering them inapplicable to genes found only in closely related sequences. Here, we present a method for the detection of selection signatures on overlapping reading frames by using closely related sequences, and apply the method to several known overlapping genes, and to an overlapping ORF on the negative strand of segment 8 of influenza A virus (NEG8), for which the suggestion has been made that it is functional. We find no evidence that NEG8 is under selection, suggesting that the intact reading frame might be non-functional, although we cannot fully exclude the possibility that the method is not sensitive enough to detect the signature of selection acting on this gene. We present the limitations of the method using known overlapping genes and suggest several approaches to improve it in future studies. Finally, we examine alternative explanations for the sequence conservation of NEG8 in the absence of selection. We show that overlap type and genomic context affect the conservation of intact overlapping ORFs and should therefore be considered in any attempt of estimating the signature of selection in overlapping genes.

RNA in pieces

Eukaryotic genomes accommodate numerous types of information within diverse DNA and RNA sequence elements. At many loci, these elements overlap and the same sequence is read multiple times during the production, processing, localization, function and turnover of a single transcript. Moreover, two or more transcripts from the same locus might use a common sequence in different ways, to perform distinct biological roles. Recent results show that many transcripts also undergo post-transcriptional cleavage to release specific fragments, which can then function independently. This phenomenon appears remarkably widespread, with even well-documented transcript classes such as messenger RNAs yielding fragments. RNA fragmentation significantly expands the already extraordinary spectrum of transcripts present within eukaryotic cells, and also calls into question how the 'gene' should be defined.

Detection of functional overlapping genes: simulation and case studies

As far as protein-coding genes are concerned, there is a non-zero probability that at least one of the five possible overlapping sequences of any gene will contain an open-reading frame (ORF) of a length that may be suitable for coding a functional protein. It is, however, very difficult to determine whether or not such an ORF is functional. Recently, we proposed a method that predicts functionality of an overlapping ORF if it can be shown that it has been subject to purifying selection during its evolution. Here, we use simulation to test this method under several conditions and compare it with the method of Firth and Brown. We found that under most conditions, our method detects functional overlapping genes with higher sensitivity than Firth and Brown's method, while maintaining high specificity. Further, we tested the hypothesis that the two aminoacyl tRNA synthetase classes have originated from a pair of overlapping genes. A central piece of evidence ostensibly supporting this hypothesis is the assertion that an overlapping ORF of a heat-shock protein-70 gene, which exhibits some similarity to class 2 aminoacyl tRNA synthetases, is functional. We found signature of purifying selection only in highly divergent sequences, suggesting that the method yields false-positives in high sequence divergence and that the overlapping ORF is not a functional gene. Finally, we examined three cases of overlap in the human genome. We find varying signatures of purifying selection acting on these overlaps, raising the possibility that two of the overlapping genes may not be functional.

Functional impact of transposable elements using bioinformatic analysis and a comparative genomic approach

A dual coding event, which is the translation of different isoforms from a single gene, is one of the special patterns among the alternative splicing events. This is an important mechanism for the regulation of protein diversity in human and mouse genomes. Although the regulation for dual coding events has been characterized in a few genes, the individual mechanism remains unclear. Numerous studies have described the exonization of transposable elements, which is the splicing mediated insertion of transposable element sequence fragments into mature mRNAs. Therefore, in this study, we investigated the number of transposable element (TE)-derived dual coding genes in human, chimpanzee and mouse genomes. TE fusion exons appeared in the dual coding regions of 309 human genes. Functional protein domain alterations by TE-derived dual coding events were observed in 129 human genes. Comparative TE-derived dual coding events were also analyzed in chimpanzee and mouse orthologs. Seventy chimpanzee orthologs had TE-derived dual coding events, but mouse orthologs did not have any TE-derived dual coding events. Taken together, our analyses listed the number of TE-derived dual coding genes which could be investigated by experimental analysis and suggested that TE-derived dual coding events were major sources for the functional diversity of human genes, but not mouse genes.