Prediction of mRNA polyadenylation sites by support vector machine

The emerging theme of 3'UTR mRNA isoform regulation in reprogramming of cell metabolism

The 3' untranslated region (3'UTR) of mRNA plays a key role in the post-transcriptional regulation of gene expression. Most eukaryotic protein-coding genes express 3'UTR isoforms owing to alternative cleavage and polyadenylation (APA). The 3'UTR isoform expression profile of a cell changes in cell proliferation, differentiation, and stress conditions. Here, we review the emerging theme of regulation of 3'UTR isoforms in cell metabolic reprogramming, focusing on cell growth and autophagy responses through the mTOR pathway. We discuss regulatory events that converge on the Cleavage Factor I complex, a master regulator of APA in 3'UTRs, and recent understandings of isoform-specific m6A modification and endomembrane association in determining differential metabolic fates of 3'UTR isoforms.

A Survey on Methods for Predicting Polyadenylation Sites from DNA Sequences, Bulk RNA-seq, and Single-cell RNA-seq

Alternative polyadenylation (APA) plays important roles in modulating mRNA stability, translation, and subcellular localization, and contributes extensively to shaping eukaryotic transcriptome complexity and proteome diversity. Identification of poly(A) sites (pAs) on a genome-wide scale is a critical step toward understanding the underlying mechanism of APA-mediated gene regulation. A number of established computational tools have been proposed to predict pAs from diverse genomic data. Here we provided an exhaustive overview of computational approaches for predicting pAs from DNA sequences, bulk RNA sequencing (RNA-seq) data, and single-cell RNA sequencing (scRNA-seq) data. Particularly, we examined several representative tools using bulk RNA-seq and scRNA-seq data from peripheral blood mononuclear cells and put forward operable suggestions on how to assess the reliability of pAs predicted by different tools. We also proposed practical guidelines on choosing appropriate methods applicable to diverse scenarios. Moreover, we discussed in depth the challenges in improving the performance of pA prediction and benchmarking different methods. Additionally, we highlighted outstanding challenges and opportunities using new machine learning and integrative multi-omics techniques, and provided our perspective on how computational methodologies might evolve in the future for non-3' untranslated region, tissue-specific, cross-species, and single-cell pA prediction.

From shallow to deep: some lessons learned from application of machine learning for recognition of functional genomic elements in human genome

Identification of genomic signals as indicators for functional genomic elements is one of the areas that received early and widespread application of machine learning methods. With time, the methods applied grew in variety and generally exhibited a tendency to improve their ability to identify some major genomic and transcriptomics signals. The evolution of machine learning in genomics followed a similar path to applications of machine learning in other fields. These were impacted in a major way by three dominant developments, namely an enormous increase in availability and quality of data, a significant increase in computational power available to machine learning applications, and finally, new machine learning paradigms, of which deep learning is the most well-known example. It is not easy in general to distinguish factors leading to improvements in results of applications of machine learning. This is even more so in the field of genomics, where the advent of next-generation sequencing and the increased ability to perform functional analysis of raw data have had a major effect on the applicability of machine learning in OMICS fields. In this paper, we survey the results from a subset of published work in application of machine learning in the recognition of genomic signals and regions in human genome and summarize some lessons learnt from this endeavor. There is no doubt that a significant progress has been made both in terms of accuracy and reliability of models. Questions remain however whether the progress has been sufficient and what these developments bring to the field of genomics in general and human genomics in particular. Improving usability, interpretability and accuracy of models remains an important open challenge for current and future research in application of machine learning and more generally of artificial intelligence methods in genomics.

In-Depth Annotation of the Drosophila Bithorax-Complex Reveals the Presence of Several Alternative ORFs That Could Encode for Motif-Rich Peptides

It is recognized that a large proportion of eukaryotic RNAs and proteins is not produced from conventional genes but from short and alternative (alt) open reading frames (ORFs) that are not captured by gene prediction programs. Here we present an in silico prediction of altORFs by applying several selecting filters based on evolutionary conservation and annotations of previously characterized altORF peptides. Our work was performed in the Bithorax-complex (BX-C), which was one of the first genomic regions described to contain long non-coding RNAs in Drosophila. We showed that several altORFs could be predicted from coding and non-coding sequences of BX-C. In addition, the selected altORFs encode for proteins that contain several interesting molecular features, such as the presence of transmembrane helices or a general propensity to be rich in short interaction motifs. Of particular interest, one altORF encodes for a protein that contains a peptide sequence found in specific isoforms of two Drosophila Hox proteins. Our work thus suggests that several altORF proteins could be produced from a particular genomic region known for its critical role during Drosophila embryonic development. The molecular signatures of these altORF proteins further suggests that several of them could make numerous protein–protein interactions and be of functional importance in vivo.

AI applications in functional genomics

We review the current applications of artificial intelligence (AI) in functional genomics. The recent explosion of AI follows the remarkable achievements made possible by "deep learning", along with a burst of "big data" that can meet its hunger. Biology is about to overthrow astronomy as the paradigmatic representative of big data producer. This has been made possible by huge advancements in the field of high throughput technologies, applied to determine how the individual components of a biological system work together to accomplish different processes. The disciplines contributing to this bulk of data are collectively known as functional genomics. They consist in studies of: i) the information contained in the DNA (genomics); ii) the modifications that DNA can reversibly undergo (epigenomics); iii) the RNA transcripts originated by a genome (transcriptomics); iv) the ensemble of chemical modifications decorating different types of RNA transcripts (epitranscriptomics); v) the products of protein-coding transcripts (proteomics); and vi) the small molecules produced from cell metabolism (metabolomics) present in an organism or system at a given time, in physiological or pathological conditions. After reviewing main applications of AI in functional genomics, we discuss important accompanying issues, including ethical, legal and economic issues and the importance of explainability.

MicroSalmon: A Comprehensive, Searchable Resource of Predicted MicroRNA Targets and 3'UTR Cis-Regulatory Elements in the Full-Length Sequenced Atlantic Salmon Transcriptome

Complete 3′UTRs unambiguously assigned to specific mRNA isoforms from the Atlantic salmon full-length (FL) transcriptome were collected into a 3′UTRome. miRNA response elements (MREs) and other cis-regulatory motifs were subsequently predicted and assigned to 3′UTRs of all FL-transcripts. The MicroSalmon GitHub repository provides all results. RNAHybrid and sRNAtoolbox tools predicted the MREs. UTRscan and the Teiresias algorithm predicted other 3′UTR cis-acting motifs, both known vertebrate motifs and putative novel motifs. MicroSalmon provides search programs to retrieve all FL-transcripts targeted by a miRNA (median number 1487), all miRNAs targeting an FL-transcript (median number 27), and other cis-acting motifs. As thousands of FL-transcripts may be targets of each miRNA, additional experimental strategies are necessary to reduce the likely true and relevant targets to a number that may be functionally validated. Low-complexity motifs known to affect mRNA decay in vertebrates were over-represented. Many of these were enriched in the terminal end, while purine- or pyrimidine-rich motifs with unknown functions were enriched immediately downstream of the stop codon. Furthermore, several novel complex motifs were over-represented, indicating conservation and putative function. In conclusion, MicroSalmon is an extensive and useful, searchable resource for study of Atlantic salmon transcript regulation by miRNAs and cis-acting 3′UTR motifs.

Discovery of widespread transcription initiation at microsatellites predictable by sequence-based deep neural network

Using the Cap Analysis of Gene Expression (CAGE) technology, the FANTOM5 consortium provided one of the most comprehensive maps of transcription start sites (TSSs) in several species. Strikingly, ~72% of them could not be assigned to a specific gene and initiate at unconventional regions, outside promoters or enhancers. Here, we probe these unassigned TSSs and show that, in all species studied, a significant fraction of CAGE peaks initiate at microsatellites, also called short tandem repeats (STRs). To confirm this transcription, we develop Cap Trap RNA-seq, a technology which combines cap trapping and long read MinION sequencing. We train sequence-based deep learning models able to predict CAGE signal at STRs with high accuracy. These models unveil the importance of STR surrounding sequences not only to distinguish STR classes, but also to predict the level of transcription initiation. Importantly, genetic variants linked to human diseases are preferentially found at STRs with high transcription initiation level, supporting the biological and clinical relevance of transcription initiation at STRs. Together, our results extend the repertoire of non-coding transcription associated with DNA tandem repeats and complexify STR polymorphism.

Generation of 3'UTR knockout cell lines by CRISPR/Cas9-mediated genome editing

In addition to the protein code, messenger RNAs (mRNAs) also contain untranslated regions (UTRs). 3'UTRs span the region between the translational stop codon and the poly(A) tail. Sequence elements located in 3'UTRs are essential for pre-mRNA processing. 3'UTRs also contain elements that can regulate protein abundance, localization, and function. At least half of all human genes use alternative cleavage and polyadenylation (APA) to further diversify the regulatory potential of protein functions. Traditional gene editing approaches are designed to disrupt the production of functional proteins. Here, we describe a method that allows investigators to manipulate 3'UTR sequences of endogenous genes for both single- 3'UTR and multi-3'UTR genes. As 3'UTRs can regulate individual functions of proteins, techniques to manipulate 3'UTRs at endogenous gene loci will help to disentangle multi-functionality of proteins. Furthermore, the ability to directly examine the impact of gene regulatory elements in 3'UTRs will provide further insights into their functional significance.

Advances in the Bioinformatics Knowledge of mRNA Polyadenylation in Baculovirus Genes

Baculoviruses are a group of insect viruses with large circular dsDNA genomes exploited in numerous biotechnological applications, such as the biological control of agricultural pests, the expression of recombinant proteins or the gene delivery of therapeutic sequences in mammals, among others. Their genomes encode between 80 and 200 proteins, of which 38 are shared by all reported species. Thanks to multi-omic studies, there is remarkable information about the baculoviral proteome and the temporality in the virus gene expression. This allows some functional elements of the genome to be very well described, such as promoters and open reading frames. However, less information is available about the transcription termination signals and, consequently, there are still imprecisions about what are the limits of the transcriptional units present in the baculovirus genomes and how is the processing of the 3′ end of viral mRNA. Regarding to this, in this review we provide an update about the characteristics of DNA signals involved in this process and we contribute to their correct prediction through an exhaustive analysis that involves bibliography information, data mining, RNA structure and a comprehensive study of the core gene 3′ ends from 180 baculovirus genomes.

Poly(A)-DG: A deep-learning-based domain generalization method to identify cross-species Poly(A) signal without prior knowledge from target species

In eukaryotes, polyadenylation (poly(A)) is an essential process during mRNA maturation. Identifying the cis-determinants of poly(A) signal (PAS) on the DNA sequence is the key to understand the mechanism of translation regulation and mRNA metabolism. Although machine learning methods were widely used in computationally identifying PAS, the need for tremendous amounts of annotation data hinder applications of existing methods in species without experimental data on PAS. Therefore, cross-species PAS identification, which enables the possibility to predict PAS from untrained species, naturally becomes a promising direction. In our works, we propose a novel deep learning method named Poly(A)-DG for cross-species PAS identification. Poly(A)-DG consists of a Convolution Neural Network-Multilayer Perceptron (CNN-MLP) network and a domain generalization technique. It learns PAS patterns from the training species and identifies PAS in target species without re-training. To test our method, we use four species and build cross-species training sets with two of them and evaluate the performance of the remaining ones. Moreover, we test our method against insufficient data and imbalanced data issues and demonstrate that Poly(A)-DG not only outperforms state-of-the-art methods but also maintains relatively high accuracy when it comes to a smaller or imbalanced training set.

Elements at the 5' end of Xist harbor SPEN-independent transcriptional antiterminator activity

The Xist lncRNA requires Repeat A, a conserved RNA element located in its 5' end, to induce gene silencing during X-chromosome inactivation. Intriguingly, Repeat A is also required for production of Xist. While silencing by Repeat A requires the protein SPEN, how Repeat A promotes Xist production remains unclear. We report that in mouse embryonic stem cells, expression of a transgene comprising the first two kilobases of Xist (Xist-2kb) causes transcriptional readthrough of downstream polyadenylation sequences. Readthrough required Repeat A and the ∼750 nucleotides downstream, did not require SPEN, and was attenuated by splicing. Despite associating with SPEN and chromatin, Xist-2kb did not robustly silence transcription, whereas a 5.5-kb Xist transgene robustly silenced transcription and read through its polyadenylation sequence. Longer, spliced Xist transgenes also induced robust silencing yet terminated efficiently. Thus, in contexts examined here, Xist requires sequence elements beyond its first two kilobases to robustly silence transcription, and the 5' end of Xist harbors SPEN-independent transcriptional antiterminator activity that can repress proximal cleavage and polyadenylation. In endogenous contexts, this antiterminator activity may help produce full-length Xist RNA while rendering the Xist locus resistant to silencing by the same repressive complexes that the lncRNA recruits to other genes.

SANPolyA: a deep learning method for identifying Poly(A) signals

MOTIVATION

Polyadenylation plays a regulatory role in transcription. The recognition of polyadenylation signal (PAS) motif sequence is an important step in polyadenylation. In the past few years, some statistical machine learning-based and deep learning-based methods have been proposed for PAS identification. Although these methods predict PAS with success, there is room for their improvement on PAS identification.

RESULTS

In this study, we proposed a deep neural network-based computational method, called SANPolyA, for identifying PAS in human and mouse genomes. SANPolyA requires no manually crafted sequence features. We compared our method SANPolyA with several previous PAS identification methods on several PAS benchmark datasets. Our results showed that SANPolyA outperforms the state-of-art methods. SANPolyA also showed good performance on leave-one-motif-out evaluation.

AVAILABILITY AND IMPLEMENTATION

https://github.com/yuht4/SANPolyA.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Systematic identification of functional SNPs interrupting 3'UTR polyadenylation signals

Alternative polyadenylation (APA) is emerging as a widespread regulatory layer since the majority of human protein-coding genes contain several polyadenylation (p(A)) sites in their 3’UTRs. By generating isoforms with different 3’UTR length, APA potentially affects mRNA stability, translation efficiency, nuclear export, and cellular localization. Polyadenylation sites are regulated by adjacent RNA cis-regulatory elements, the principals among them are the polyadenylation signal (PAS) AAUAAA and its main variant AUUAAA, typically located ~20-nt upstream of the p(A) site. Mutations in PAS and other auxiliary poly(A) cis-elements in the 3’UTR of several genes have been shown to cause human Mendelian diseases, and to date, only a few common SNPs that regulate APA were associated with complex diseases. Here, we systematically searched for SNPs that affect gene expression and human traits by modulation of 3’UTR APA. First, focusing on the variants most likely to exert the strongest effect, we identified 2,305 SNPs that interrupt the canonical PAS or its main variant. Implementing pA-QTL tests using GTEx RNA-seq data, we identified 330 PAS SNPs (called PAS pA-QTLs) that were significantly associated with the usage of their p(A) site. As expected, PAS-interrupting alleles were mostly linked with decreased cleavage at their p(A) site and the consequential 3’UTR lengthening. However, interestingly, in ~10% of the cases, the PAS-interrupting allele was associated with increased usage of an upstream p(A) site and 3’UTR shortening. As an indication of the functional effects of these PAS pA-QTLs on gene expression and complex human traits, we observed for few dozens of them marked colocalization with eQTL and/or GWAS signals. The PAS-interrupting alleles linked with 3’UTR lengthening were also strongly associated with decreased gene expression, indicating that shorter isoforms generated by APA are generally more stable than longer ones. Last, we carried out an extended, genome-wide analysis of 3’UTR variants and detected thousands of additional pA-QTLs having weaker effects compared to the PAS pA-QTLs.

mRNA molecules that encode for proteins end with a long stretch of adenosines, called poly(A) tail. The poly(A) tail contributes to the stability of the mRNA molecules, their translation to proteins and their import from the nucleus to the cytoplasm. The process of adding this tail to the mRNAs is called polyadenylation, and the termination site on the mRNAs at which the poly(A) tail is added is called the poly(A) site. In recent years it became evident that the vast majority of mRNAs of human genes contain several alternative poly(A) sites and their usage generates different mRNA isoforms that differ in their stability and translation efficiency. Therefore, alternative polyadenylation (APA) is emerging as a novel and important, yet underexplored, mechanism that regulate gene expression. The choice between alternative p(A) sites in an mRNA molecule is regulated by regulatory sequences located within a region in the mRNA called the 3’ untranslated region (3’UTR). A major challenge in present human genetics research is to understand how common genetic variants affect individuals’ health. In our study, we systematically identified dozens of genetic variants that affect the choice between alternative p(A) sites and demonstrated that by that, these variants influence the expression level of the target genes. Our results help to illuminate a novel mechanism by which genetic variants that are common in the population affect different traits including our risk for developing diseases.

DeepPASTA: deep neural network based polyadenylation site analysis

MOTIVATION

Alternative polyadenylation (polyA) sites near the 3' end of a pre-mRNA create multiple mRNA transcripts with different 3' untranslated regions (3' UTRs). The sequence elements of a 3' UTR are essential for many biological activities such as mRNA stability, sub-cellular localization, protein translation, protein binding and translation efficiency. Moreover, numerous studies in the literature have reported the correlation between diseases and the shortening (or lengthening) of 3' UTRs. As alternative polyA sites are common in mammalian genes, several machine learning tools have been published for predicting polyA sites from sequence data. These tools either consider limited sequence features or use relatively old algorithms for polyA site prediction. Moreover, none of the previous tools consider RNA secondary structures as a feature to predict polyA sites.

RESULTS

In this paper, we propose a new deep learning model, called DeepPASTA, for predicting polyA sites from both sequence and RNA secondary structure data. The model is then extended to predict tissue-specific polyA sites. Moreover, the tool can predict the most dominant (i.e. frequently used) polyA site of a gene in a specific tissue and relative dominance when two polyA sites of the same gene are given. Our extensive experiments demonstrate that DeepPASTA signisficantly outperforms the existing tools for polyA site prediction and tissue-specific relative and absolute dominant polyA site prediction.

AVAILABILITY AND IMPLEMENTATION

https://github.com/arefeen/DeepPASTA.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Emerging Roles of RNA 3'-end Cleavage and Polyadenylation in Pathogenesis, Diagnosis and Therapy of Human Disorders

A crucial feature of gene expression involves RNA processing to produce 3′ ends through a process termed 3′ end cleavage and polyadenylation (CPA). This ensures the nascent RNA molecule can exit the nucleus and be translated to ultimately give rise to a protein which can execute a function. Further, alternative polyadenylation (APA) can produce distinct transcript isoforms, profoundly expanding the complexity of the transcriptome. CPA is carried out by multi-component protein complexes interacting with multiple RNA motifs and is tightly coupled to transcription, other steps of RNA processing, and even epigenetic modifications. CPA and APA contribute to the maintenance of a multitude of diverse physiological processes. It is therefore not surprising that disruptions of CPA and APA can lead to devastating disorders. Here, we review potential CPA and APA mechanisms involving both loss and gain of function that can have tremendous impacts on health and disease. Ultimately we highlight the emerging diagnostic and therapeutic potential CPA and APA offer.

DeeReCT-PolyA: a robust and generic deep learning method for PAS identification

Motivation

Polyadenylation is a critical step for gene expression regulation during the maturation of mRNA. An accurate and robust method for poly(A) signals (PASs) identification is not only desired for the purpose of better transcripts’ end annotation, but can also help us gain a deeper insight of the underlying regulatory mechanism. Although many methods have been proposed for PAS recognition, most of them are PAS motif- and human-specific, which leads to high risks of overfitting, low generalization power, and inability to reveal the connections between the underlying mechanisms of different mammals.

Results

In this work, we propose a robust, PAS motif agnostic, and highly interpretable and transferrable deep learning model for accurate PAS recognition, which requires no prior knowledge or human-designed features. We show that our single model trained over all human PAS motifs not only outperforms the state-of-the-art methods trained on specific motifs, but can also be generalized well to two mouse datasets. Moreover, we further increase the prediction accuracy by transferring the deep learning model trained on the data of one species to the data of a different species. Several novel underlying poly(A) patterns are revealed through the visualization of important oligomers and positions in our trained models. Finally, we interpret the deep learning models by converting the convolutional filters into sequence logos and quantitatively compare the sequence logos between human and mouse datasets.

Availability and implementation

https://github.com/likesum/DeeReCT-PolyA

Supplementary information

Supplementary data are available at Bioinformatics online.

Plant 3' Regulatory Regions From mRNA-Encoding Genes and Their Uses to Modulate Expression

Molecular biotechnology has made it possible to explore the potential of plants for different purposes. The 3' regulatory regions have a great diversity of cis-regulatory elements directly involved in polyadenylation, stability, transport and mRNA translation, essential to achieve the desired levels of gene expression. A complex interaction between the cleavage and polyadenylation molecular complex and cis-elements determine the polyadenylation site, which may result in the choice of non-canonical sites, resulting in alternative polyadenylation events, involved in the regulation of more than 80% of the genes expressed in plants. In addition, after transcription, a wide array of RNA-binding proteins interacts with cis-acting elements located mainly in the 3' untranslated region, determining the fate of mRNAs in eukaryotic cells. Although a small number of 3' regulatory regions have been identified and validated so far, many studies have shown that plant 3' regulatory regions have a higher potential to regulate gene expression in plants compared to widely used 3' regulatory regions, such as NOS and OCS from Agrobacterium tumefaciens and 35S from cauliflower mosaic virus. In this review, we discuss the role of 3' regulatory regions in gene expression, and the superior potential that plant 3' regulatory regions have compared to NOS, OCS and 35S 3' regulatory regions.

Cell-type-specific analysis of alternative polyadenylation using single-cell transcriptomics data

Alternative polyadenylation (APA) is emerging as an important layer of gene regulation because the majority of mammalian protein-coding genes contain multiple polyadenylation (pA) sites in their 3' UTR. By alteration of 3' UTR length, APA can considerably affect post-transcriptional gene regulation. Yet, our understanding of APA remains rudimentary. Novel single-cell RNA sequencing (scRNA-seq) techniques allow molecular characterization of different cell types to an unprecedented degree. Notably, the most popular scRNA-seq protocols specifically sequence the 3' end of transcripts. Building on this property, we implemented a method for analysing patterns of APA regulation from such data. Analyzing multiple datasets from diverse tissues, we identified widespread modulation of APA in different cell types resulting in global 3' UTR shortening/lengthening and enhanced cleavage at intronic pA sites. Our results provide a proof-of-concept demonstration that the huge volume of scRNA-seq data that accumulates in the public domain offers a unique resource for the exploration of APA based on a very broad collection of cell types and biological conditions.

Alternative polyadenylation of mRNA and its role in cancer

Alternative polyadenylation (APA) is a molecular process that generates diversity at the 3′ end of RNA polymerase II transcripts from over 60% of human genes. APA is derived from the existence of multiple polyadenylation signals (PAS) within the same transcript, and results in the differential inclusion of sequence information at the 3′ end. While APA can occur between two PASs allowing for generation of transcripts with distinct coding potential from a single gene, most APA occurs within the untranslated region (3′UTR) and changes the length and content of these non-coding sequences. APA within the 3′UTR can have tremendous impact on its regulatory potential of the mRNA through a variety of mechanisms, and indeed this layer of gene expression regulation has profound impact on processes vital to cell growth and development. Recent studies have particularly highlighted the importance of APA dysregulation in cancer onset and progression. Here, we review the current knowledge of APA and its impacts on mRNA stability, translation, localization and protein localization. We also discuss the implications of APA dysregulation in cancer research and therapy.

Inference of the human polyadenylation code

Motivation

Processing of transcripts at the 3′-end involves cleavage at a polyadenylation site followed by the addition of a poly(A)-tail. By selecting which site is cleaved, the process of alternative polyadenylation enables genes to produce transcript isoforms with different 3′-ends. To facilitate the identification and treatment of disease-causing mutations that affect polyadenylation and to understand the sequence determinants underlying this regulatory process, a computational model that can accurately predict polyadenylation patterns from genomic features is desirable.

Results

Previous works have focused on identifying candidate polyadenylation sites and classifying tissue-specific sites. By training on how multiple sites in genes are competitively selected for polyadenylation from 3′-end sequencing data, we developed a deep learning model that can predict the tissue-specific strength of a polyadenylation site in the 3′ untranslated region of the human genome given only its genomic sequence. We demonstrate the model’s broad utility on multiple tasks, without any application-specific training. The model can be used to predict which polyadenylation site is more likely to be selected in genes with multiple sites. It can be used to scan the 3′ untranslated region to find candidate polyadenylation sites. It can be used to classify the pathogenicity of variants near annotated polyadenylation sites in ClinVar. It can also be used to anticipate the effect of antisense oligonucleotide experiments to redirect polyadenylation. We provide analysis on how different features affect the model’s predictive performance and a method to identify sensitive regions of the genome at the single-based resolution that can affect polyadenylation regulation.

Supplementary information

Supplementary data are available at Bioinformatics online.

Sequence determinants of polyadenylation-mediated regulation

The cleavage and polyadenylation reaction is a crucial step in transcription termination and pre-mRNA maturation in human cells. Despite extensive research, the encoding of polyadenylation-mediated regulation of gene expression within the DNA sequence is not well understood. Here, we utilized a massively parallel reporter assay to inspect the effect of over 12,000 rationally designed polyadenylation sequences (PASs) on reporter gene expression and cleavage efficiency. We find that the PAS sequence can modulate gene expression by over five orders of magnitude. By using a uniquely designed scanning mutagenesis data set, we gain mechanistic insight into various modes of action by which the cleavage efficiency affects the sensitivity or robustness of the PAS to mutation. Furthermore, we employ motif discovery to identify both known and novel sequence motifs associated with PAS-mediated regulation. By leveraging the large scale of our data, we train a deep learning model for the highly accurate prediction of RNA levels from DNA sequence alone (R = 0.83). Moreover, we devise unique approaches for predicting exact cleavage sites for our reporter constructs and for endogenous transcripts. Taken together, our results expand our understanding of PAS-mediated regulation, and provide an unprecedented resource for analyzing and predicting PAS for regulatory genomics applications.

Atlas of Subcellular RNA Localization Revealed by APEX-Seq

We introduce APEX-seq, a method for RNA sequencing based on direct proximity labeling of RNA using the peroxidase enzyme APEX2. APEX-seq in nine distinct subcellular locales produced a nanometer-resolution spatial map of the human transcriptome as a resource, revealing extensive patterns of localization for diverse RNA classes and transcript isoforms. We uncover a radial organization of the nuclear transcriptome, which is gated at the inner surface of the nuclear pore for cytoplasmic export of processed transcripts. We identify two distinct pathways of messenger RNA localization to mitochondria, each associated with specific sets of transcripts for building complementary macromolecular machines within the organelle. APEX-seq should be widely applicable to many systems, enabling comprehensive investigations of the spatial transcriptome.

A Deep Neural Network for Predicting and Engineering Alternative Polyadenylation

Alternative polyadenylation (APA) is a major driver of transcriptome diversity in human cells. Here, we use deep learning to predict APA from DNA sequence alone. We trained our model (APARENT, APA REgression NeT) on isoform expression data from over 3 million APA reporters. APARENT's predictions are highly accurate when tasked with inferring APA in synthetic and human 3'UTRs. Visualizing features learned across all network layers reveals that APARENT recognizes sequence motifs known to recruit APA regulators, discovers previously unknown sequence determinants of 3' end processing, and integrates these features into a comprehensive, interpretable, cis-regulatory code. We apply APARENT to forward engineer functional polyadenylation signals with precisely defined cleavage position and isoform usage and validate predictions experimentally. Finally, we use APARENT to quantify the impact of genetic variants on APA. Our approach detects pathogenic variants in a wide range of disease contexts, expanding our understanding of the genetic origins of disease.

The Untranslated Regions of mRNAs in Cancer

The 5' and 3' untranslated regions (UTRs) regulate crucial aspects of post-transcriptional gene regulation that are necessary for the maintenance of cellular homeostasis. When these processes go awry through mutation or misexpression of certain regulatory elements, the subsequent deregulation of oncogenic gene expression can drive or enhance cancer pathogenesis. Although the number of known cancer-related mutations in UTR regulatory elements has recently increased markedly as a result of advances in whole-genome sequencing, little is known about how the majority of these genetic aberrations contribute functionally to disease. In this review we explore the regulatory functions of UTRs, how they are co-opted in cancer, new technologies to interrogate cancerous UTRs, and potential therapeutic opportunities stemming from these regions.

NAA10 polyadenylation signal variants cause syndromic microphthalmia

BACKGROUND

A single variant in NAA10 (c.471+2T>A), the gene encoding N-acetyltransferase 10, has been associated with Lenz microphthalmia syndrome. In this study, we aimed to identify causative variants in families with syndromic X-linked microphthalmia.

METHODS

Three families, including 15 affected individuals with syndromic X-linked microphthalmia, underwent analyses including linkage analysis, exome sequencing and targeted gene sequencing. The consequences of two identified variants in NAA10 were evaluated using quantitative PCR and RNAseq.

RESULTS

Genetic linkage analysis in family 1 supported a candidate region on Xq27-q28, which included NAA10. Exome sequencing identified a hemizygous NAA10 polyadenylation signal (PAS) variant, chrX:153,195,397T>C, c.*43A>G, which segregated with the disease. Targeted sequencing of affected males from families 2 and 3 identified distinct NAA10 PAS variants, chrX:g.153,195,401T>C, c.*39A>G and chrX:g.153,195,400T>C, c.*40A>G. All three variants were absent from gnomAD. Quantitative PCR and RNAseq showed reduced NAA10 mRNA levels and abnormal 3' UTRs in affected individuals. Targeted sequencing of NAA10 in 376 additional affected individuals failed to identify variants in the PAS.

CONCLUSION

These data show that PAS variants are the most common variant type in NAA10-associated syndromic microphthalmia, suggesting reduced RNA is the molecular mechanism by which these alterations cause microphthalmia/anophthalmia. We reviewed recognised variants in PAS associated with Mendelian disorders and identified only 23 others, indicating that NAA10 harbours more than 10% of all known PAS variants. We hypothesise that PAS in other genes harbour unrecognised pathogenic variants associated with Mendelian disorders. The systematic interrogation of PAS could improve genetic testing yields.

A compendium of conserved cleavage and polyadenylation events in mammalian genes

Cleavage and polyadenylation is essential for 3' end processing of almost all eukaryotic mRNAs. Recent studies have shown widespread alternative cleavage and polyadenylation (APA) events leading to mRNA isoforms with different 3' UTRs and/or coding sequences. Here, we present a compendium of conserved cleavage and polyadenylation sites (PASs) in mammalian genes, based on approximately 1.2 billion 3' end sequencing reads from more than 360 human, mouse, and rat samples. We show that ∼80% of mammalian mRNA genes contain at least one conserved PAS, and ∼50% have conserved APA events. PAS conservation generally reduces promiscuous 3' end processing, stabilizing gene expression levels across species. Conservation of APA correlates with gene age, gene expression features, and gene functions. Genes with certain functions, such as cell morphology, cell proliferation, and mRNA metabolism, are particularly enriched with conserved APA events. Whereas tissue-specific genes typically have a low APA rate, brain-specific genes tend to evolve APA. In addition, we show enrichment of mRNA destabilizing motifs in alternative 3' UTR sequences, leading to substantial differences in mRNA stability between 3' UTR isoforms. Using conserved PASs, we reveal sequence motifs surrounding APA sites and a preference of adenosine at the cleavage site. Furthermore, we show that mutations of U-rich motifs around the PAS often accompany APA profile differences between species. Analysis of lncRNA PASs indicates a mechanism of PAS fixation through evolution of A-rich motifs. Taken together, our results present a comprehensive view of PAS evolution in mammals, and a phylogenic perspective on APA functions.

QAPA: a new method for the systematic analysis of alternative polyadenylation from RNA-seq data

Alternative polyadenylation (APA) affects most mammalian genes. The genome-wide investigation of APA has been hampered by an inability to reliably profile it using conventional RNA-seq. We describe 'Quantification of APA' (QAPA), a method that infers APA from conventional RNA-seq data. QAPA is faster and more sensitive than other methods. Application of QAPA reveals discrete, temporally coordinated APA programs during neurogenesis and that there is little overlap between genes regulated by alternative splicing and those by APA. Modeling of these data uncovers an APA sequence code. QAPA thus enables the discovery and characterization of programs of regulated APA using conventional RNA-seq.

Untranslated Parts of Genes Interpreted: Making Heads or Tails of High-Throughput Transcriptomic Data via Computational Methods: Computational methods to discover and quantify isoforms with alternative untranslated regions

In this review we highlight the importance of defining the untranslated parts of transcripts, and present a number of computational approaches for the discovery and quantification of alternative transcription start and poly-adenylation events in high-throughput transcriptomic data. The fate of eukaryotic transcripts is closely linked to their untranslated regions, which are determined by the position at which transcription starts and ends at a genomic locus. Although the extent of alternative transcription starts and alternative poly-adenylation sites has been revealed by sequencing methods focused on the ends of transcripts, the application of these methods is not yet widely adopted by the community. We suggest that computational methods applied to standard high-throughput technologies are a useful, albeit less accurate, alternative to the expertise-demanding 5' and 3' sequencing and they are the only option for analysing legacy transcriptomic data. We review these methods here, focusing on technical challenges and arguing for the need to include better normalization of the data and more appropriate statistical models of the expected variation in the signal.

Comparative analysis of alternative polyadenylation in S. cerevisiae and S. pombe

Alternative polyadenylation (APA) is a widespread mechanism that generates mRNA isoforms with distinct properties. Here we have systematically mapped and compared cleavage and polyadenylation sites (PASs) in two yeast species, S. cerevisiae and S. pombe. Although >80% of the mRNA genes in each species were found to display APA, S. pombe showed greater 3′ UTR size differences among APA isoforms than did S. cerevisiae. PASs in different locations of gene are surrounded with distinct sequences in both species and are often associated with motifs involved in the Nrd1-Nab3-Sen1 termination pathway. In S. pombe, strong motifs surrounding distal PASs lead to higher abundances of long 3′ UTR isoforms than short ones, a feature that is opposite in S. cerevisiae. Differences in PAS placement between convergent genes lead to starkly different antisense transcript landscapes between budding and fission yeasts. In both species, short 3′ UTR isoforms are more likely to be expressed when cells are growing in nutrient-rich media, although different gene groups are affected in each species. Significantly, 3′ UTR shortening in S. pombe coordinates with up-regulation of expression for genes involved in translation during cell proliferation. Using S. pombe strains deficient for Pcf11 or Pab2, we show that reduced expression of 3′-end processing factors lengthens 3′ UTR, with Pcf11 having a more potent effect than Pab2. Taken together, our data indicate that APA mechanisms in S. pombe and S. cerevisiae are largely different: S. pombe has many of the APA features of higher species, and Pab2 in S. pombe has a different role in APA regulation than its mammalian homolog, PABPN1.

Omni-PolyA: a method and tool for accurate recognition of Poly(A) signals in human genomic DNA

BACKGROUND

Polyadenylation is a critical stage of RNA processing during the formation of mature mRNA, and is present in most of the known eukaryote protein-coding transcripts and many long non-coding RNAs. The correct identification of poly(A) signals (PAS) not only helps to elucidate the 3'-end genomic boundaries of a transcribed DNA region and gene regulatory mechanisms but also gives insight into the multiple transcript isoforms resulting from alternative PAS. Although progress has been made in the in-silico prediction of genomic signals, the recognition of PAS in DNA genomic sequences remains a challenge.

RESULTS

In this study, we analyzed human genomic DNA sequences for the 12 most common PAS variants. Our analysis has identified a set of features that helps in the recognition of true PAS, which may be involved in the regulation of the polyadenylation process. The proposed features, in combination with a recognition model, resulted in a novel method and tool, Omni-PolyA. Omni-PolyA combines several machine learning techniques such as different classifiers in a tree-like decision structure and genetic algorithms for deriving a robust classification model. We performed a comparison between results obtained by state-of-the-art methods, deep neural networks, and Omni-PolyA. Results show that Omni-PolyA significantly reduced the average classification error rate by 35.37% in the prediction of the 12 considered PAS variants relative to the state-of-the-art results.

CONCLUSIONS

The results of our study demonstrate that Omni-PolyA is currently the most accurate model for the prediction of PAS in human and can serve as a useful complement to other PAS recognition methods. Omni-PolyA is publicly available as an online tool accessible at www.cbrc.kaust.edu.sa/omnipolya/ .

A computational approach for the functional classification of the epigenome

BACKGROUND

In the last decade, advanced functional genomics approaches and deep sequencing have allowed large-scale mapping of histone modifications and other epigenetic marks, highlighting functional relationships between chromatin organization and genome function. Here, we propose a novel approach to explore functional interactions between different epigenetic modifications and extract combinatorial profiles that can be used to annotate the chromatin in a finite number of functional classes. Our method is based on non-negative matrix factorization (NMF), an unsupervised learning technique originally employed to decompose high-dimensional data in a reduced number of meaningful patterns. We applied the NMF algorithm to a set of different epigenetic marks, consisting of ChIP-seq assays for multiple histone modifications, Pol II binding and chromatin accessibility assays from human H1 cells.

RESULTS

We identified a number of chromatin profiles that contain functional information and are biologically interpretable. We also observe that epigenetic profiles are characterized by specific genomic contexts and show significant association with distinct genomic features. Moreover, analysis of RNA-seq data reveals that distinct chromatin signatures correlate with the level of gene expression.

CONCLUSIONS

Overall, our study highlights the utility of NMF in studying functional relationships between different epigenetic modifications and may provide new biological insights for the interpretation of the chromatin dynamics.

Alternative polyadenylation of mRNA precursors

Alternative polyadenylation (APA) is an RNA-processing mechanism that generates distinct 3' termini on mRNAs and other RNA polymerase II transcripts. It is widespread across all eukaryotic species and is recognized as a major mechanism of gene regulation. APA exhibits tissue specificity and is important for cell proliferation and differentiation. In this Review, we discuss the roles of APA in diverse cellular processes, including mRNA metabolism, protein diversification and protein localization, and more generally in gene regulation. We also discuss the molecular mechanisms underlying APA, such as variation in the concentration of core processing factors and RNA-binding proteins, as well as transcription-based regulation.

Complex Selection on Human Polyadenylation Signals Revealed by Polymorphism and Divergence Data

Polyadenylation is a step of mRNA processing which is crucial for its expression and stability. The major polyadenylation signal (PAS) represents a nucleotide hexamer that adheres to the AATAAA consensus sequence. Over a half of human genes have multiple cleavage and polyadenylation sites, resulting in a great diversity of transcripts differing in function, stability, and translational activity. Here, we use available whole-genome human polymorphism data together with data on interspecies divergence to study the patterns of selection acting on PAS hexamers. Common variants of PAS hexamers are depleted of single nucleotide polymorphisms (SNPs), and SNPs within PAS hexamers have a reduced derived allele frequency (DAF) and increased conservation, indicating prevalent negative selection; at the same time, the SNPs that "improve" the PAS (i.e., those leading to higher cleavage efficiency) have increased DAF, compared to those that "impair" it. SNPs are rarer at PAS of "unique" polyadenylation sites (one site per gene); among alternative polyadenylation sites, at the distal PAS and at exonic PAS. Similar trends were observed in DAFs and divergence between species of placental mammals. Thus, selection permits PAS mutations mainly at redundant and/or weakly functional PAS. Nevertheless, a fraction of the SNPs at PAS hexamers likely affect gene functions; in particular, some of the observed SNPs are associated with disease.

Poly(A) code analyses reveal key determinants for tissue-specific mRNA alternative polyadenylation

mRNA alternative polyadenylation (APA) is a critical mechanism for post-transcriptional gene regulation and is often regulated in a tissue- and/or developmental stage-specific manner. An ultimate goal for the APA field has been to be able to computationally predict APA profiles under different physiological or pathological conditions. As a first step toward this goal, we have assembled a poly(A) code for predicting tissue-specific poly(A) sites (PASs). Based on a compendium of over 600 features that have known or potential roles in PAS selection, we have generated and refined a machine-learning algorithm using multiple high-throughput sequencing-based data sets of tissue-specific and constitutive PASs. This code can predict tissue-specific PASs with >85% accuracy. Importantly, by analyzing the prediction performance based on different RNA features, we found that PAS context, including the distance between alternative PASs and the relative position of a PAS within the gene, is a key feature for determining the susceptibility of a PAS to tissue-specific regulation. Our poly(A) code provides a useful tool for not only predicting tissue-specific APA regulation, but also for studying its underlying molecular mechanisms.

The enigma of ceramide synthase regulation in mammalian cells

Ceramide synthases (CerS) are key enzymes in the lipid metabolism of eukaryotic cells. Their products, ceramides (Cer), are components of cellular membranes but also mediate signaling functions in physiological processes such as proliferation, skin barrier function and cerebellar development. In pathophysiological processes such as multiple sclerosis and tumor progression, ceramide levels are altered, which can be ascribed, partly, to dysregulation of CerS gene transcription. Most publications deal with the effects of altered ceramide levels on physiological and pathophysiological processes, but the regulation of the appropriate CerS is frequently not investigated. This is insufficient for the clarification of the role of ceramides, because most ceramide species are generated by at least two CerS. The mechanisms of CerS regulation are manifold and it seems that each CerS isoform is regulated individually. For this reason, we discuss the different CerS separately in this review. From transcriptional regulation to alteration of protein activity, the possibilities to influence CerS are diverse. Furthermore, CerS are influenced by a variety of molecules including hormones and lipids. Without claiming completeness, we provide a résumé of the regulatory mechanisms for each CerS in mammalian cells and how dysregulation of these mechanisms during physiological processes may lead to pathophysiological processes.

Computational analysis of plant polyadenylation signals

Messenger RNA polyadenylation in eukaryotes marks the end of a transcript, and the process is associated with transcription termination. Increasing evidence reveals the potential of gene expression regulation through alternative polyadenylation. The site of poly(A) addition is defined by poly(A) signals reside in the transcribed pre-mRNA. To gain further insight into poly(A) signals and their functions in defining alternative polyadenylation sites that lie within different genomic regions, SignalSleuth2 was developed to extract and analyze cis-elements from a set of data with known poly(A) sites. After obtaining the sequences surrounding the poly(A) sites, exhaustive search of short sequence motifs in specified range of nucleotide sequences are performed, variable motif sizes and rank the detected motifs based on their occurrence frequencies are tallied. It also has new functions including Position-Specific Scoring Matrix (PSSM) scores calculation and multiple scanning modes. This program is powerful in revealing underline sequence motifs surrounding any target regions in a given dataset.

A novel genome-wide polyadenylation sites recognition system based on condition random field

Polyadenylation including the cleavage of pre-mRNA and addition of a stretch of adenosines to the 3'-end is an essential step of pre-mRNA processing in eukayotes. The known regulatory role of polyadenylation in mRNA localization, stability, and translation and the emerging link between poly(A) and disease states underline the necessary to fully characterize polyadenylation sites. Several artificial intelligence methods have been proposed for poly(A) sites recognition. However, these methods are suitable to small subsets of genome sequences. It is necessary to propose a method for genome-wide recognition of poly(A) sites. Recent efforts have found a lot of poly(A) related factors on DNA level. Here, we proposed a novel genome-wide poly(A) recognition method based on the Condition Random Field (CRF) by integrating multiple features. Compared with the polya_svm (the most accurate program for prediction of poly(A) sites till date), our method had a higher performance with the area under ROC curve(0.8621 versus 0.6796). The result suggests that our method is an effective method in genome wide poly(A) sites recognition.

RNA-Binding Protein Musashi1 Is a Central Regulator of Adhesion Pathways in Glioblastoma

The conserved RNA-binding protein Musashi1 (MSI1) has emerged as a key oncogenic factor in numerous solid tumors, including glioblastoma. However, its mechanism of action has not yet been established comprehensively. To identify its target genes comprehensively and determine the main routes by which it influences glioblastoma phenotypes, we conducted individual-nucleotide resolution cross-linking and immunoprecipitation (iCLIP) experiments. We confirmed that MSI1 has a preference for UAG sequences contained in a particular structural context, especially in 3' untranslated regions. Although numerous binding sites were also identified in intronic sequences, our RNA transcriptome sequencing analysis does not favor the idea that MSI1 is a major regulator of splicing in glioblastoma cells. MSI1 target mRNAs encode proteins that function in multiple pathways of cell proliferation and cell adhesion. Since these associations indicate potentially new roles for MSI1, we investigated its impact on glioblastoma cell adhesion, morphology, migration, and invasion. These processes are known to underpin the spread and relapse of glioblastoma, in contrast to other tumors where metastasis is the main driver of recurrence and progression.

EFNA3 long noncoding RNAs induced by hypoxia promote metastatic dissemination

The presence of hypoxic regions in solid tumors is an adverse prognostic factor for patient outcome. Here, we show that hypoxia induces the expression of Ephrin-A3 through a novel hypoxia-inducible factor (HIF)-mediated mechanism. In response to hypoxia, the coding EFNA3 mRNA levels remained relatively stable, but HIFs drove the expression of previously unknown long noncoding (lnc) RNAs from EFNA3 locus and these lncRNA caused Ephrin-A3 protein accumulation. Ephrins are cell surface proteins that regulate diverse biological processes by modulating cellular adhesion and repulsion. Mounting evidence implicates deregulated ephrin function in multiple aspects of tumor biology. We demonstrate that sustained expression of both Ephrin-A3 and novel EFNA3 lncRNAs increased the metastatic potential of human breast cancer cells, possibly by increasing the ability of tumor cells to extravasate from the blood vessels into surrounding tissue. In agreement, we found a strong correlation between high EFNA3 expression and shorter metastasis-free survival in breast cancer patients. Taken together, our results suggest that hypoxia could contribute to metastatic spread of breast cancer via HIF-mediated induction of EFNA3 lncRNAs and subsequent Ephrin-A3 protein accumulation.

Computational challenges, tools, and resources for analyzing co- and post-transcriptional events in high throughput

Co- and post-transcriptional regulation of gene expression is complex and multifaceted, spanning the complete RNA lifecycle from genesis to decay. High-throughput profiling of the constituent events and processes is achieved through a range of technologies that continue to expand and evolve. Fully leveraging the resulting data is nontrivial, and requires the use of computational methods and tools carefully crafted for specific data sources and often intended to probe particular biological processes. Drawing upon databases of information pre-compiled by other researchers can further elevate analyses. Within this review, we describe the major co- and post-transcriptional events in the RNA lifecycle that are amenable to high-throughput profiling. We place specific emphasis on the analysis of the resulting data, in particular the computational tools and resources available, as well as looking toward future challenges that remain to be addressed.

TBX6 null variants and a common hypomorphic allele in congenital scoliosis

BACKGROUND

Congenital scoliosis is a common type of vertebral malformation. Genetic susceptibility has been implicated in congenital scoliosis.

METHODS

We evaluated 161 Han Chinese persons with sporadic congenital scoliosis, 166 Han Chinese controls, and 2 pedigrees, family members of which had a 16p11.2 deletion, using comparative genomic hybridization, quantitative polymerase-chain-reaction analysis, and DNA sequencing. We carried out tests of replication using an additional series of 76 Han Chinese persons with congenital scoliosis and a multicenter series of 42 persons with 16p11.2 deletions.

RESULTS

We identified a total of 17 heterozygous TBX6 null mutations in the 161 persons with sporadic congenital scoliosis (11%); we did not observe any null mutations in TBX6 in 166 controls (P<3.8×10(-6)). These null alleles include copy-number variants (12 instances of a 16p11.2 deletion affecting TBX6) and single-nucleotide variants (1 nonsense and 4 frame-shift mutations). However, the discordant intrafamilial phenotypes of 16p11.2 deletion carriers suggest that heterozygous TBX6 null mutation is insufficient to cause congenital scoliosis. We went on to identify a common TBX6 haplotype as the second risk allele in all 17 carriers of TBX6 null mutations (P<1.1×10(-6)). Replication studies involving additional persons with congenital scoliosis who carried a deletion affecting TBX6 confirmed this compound inheritance model. In vitro functional assays suggested that the risk haplotype is a hypomorphic allele. Hemivertebrae are characteristic of TBX6-associated congenital scoliosis.

CONCLUSIONS

Compound inheritance of a rare null mutation and a hypomorphic allele of TBX6 accounted for up to 11% of congenital scoliosis cases in the series that we analyzed. (Funded by the National Basic Research Program of China and others.).

An improved poly(A) motifs recognition method based on decision level fusion

Polyadenylation is the process of addition of poly(A) tail to mRNA 3' ends. Identification of motifs controlling polyadenylation plays an essential role in improving genome annotation accuracy and better understanding of the mechanisms governing gene regulation. The bioinformatics methods used for poly(A) motifs recognition have demonstrated that information extracted from sequences surrounding the candidate motifs can differentiate true motifs from the false ones greatly. However, these methods depend on either domain features or string kernels. To date, methods combining information from different sources have not been found yet. Here, we proposed an improved poly(A) motifs recognition method by combing different sources based on decision level fusion. First of all, two novel prediction methods was proposed based on support vector machine (SVM): one method is achieved by using the domain-specific features and principle component analysis (PCA) method to eliminate the redundancy (PCA-SVM); the other method is based on Oligo string kernel (Oligo-SVM). Then we proposed a novel machine-learning method for poly(A) motif prediction by marrying four poly(A) motifs recognition methods, including two state-of-the-art methods (Random Forest (RF) and HMM-SVM), and two novel proposed methods (PCA-SVM and Oligo-SVM). A decision level information fusion method was employed to combine the decision values of different classifiers by applying the DS evidence theory. We evaluated our method on a comprehensive poly(A) dataset that consists of 14,740 samples on 12 variants of poly(A) motifs and 2750 samples containing none of these motifs. Our method has achieved accuracy up to 86.13%. Compared with the four classifiers, our evidence theory based method reduces the average error rate by about 30%, 27%, 26% and 16%, respectively. The experimental results suggest that the proposed method is more effective for poly(A) motif recognition.

A novel virus-inducible enhancer of the interferon-β gene with tightly linked promoter and enhancer activities

Long-range enhancers of transcription are a key component of the genomic regulatory architecture. Recent studies have identified bi-directionally transcribed RNAs emanating from these enhancers known as eRNAs. However, it remains unclear how tightly coupled eRNA production is with enhancer activity. Through our systematic search for long-range elements that interact with the interferon-β gene, a model system for studying inducible transcription, we have identified a novel enhancer, which we have named L2 that regulates the expression of interferon-β. We have demonstrated its virus-inducible enhancer activity by analyzing epigenomic profiles, transcription factor association, nascent RNA production and activity in reporter assays. This enhancer exhibits intimately linked virus-inducible enhancer and bidirectional promoter activity that is largely dependent on a conserved Interferon Stimulated Response Element and robustly generates virus inducible eRNAs. Notably, its enhancer and promoter activities are fully retained in reporter assays even upon a complete elimination of its associated eRNA sequences. Finally, we show that L2 regulates IFNB1 expression by siRNA knockdown of eRNAs, and the deletion of L2 in a BAC transfection assay. Thus, L2 is a novel enhancer that regulates IFNB1 and whose eRNAs exert significant activity in vivo that is distinct from those activities recapitulated in the luciferase reporter assays.

Fip1 regulates mRNA alternative polyadenylation to promote stem cell self-renewal

mRNA alternative polyadenylation (APA) plays a critical role in post-transcriptional gene control and is highly regulated during development and disease. However, the regulatory mechanisms and functional consequences of APA remain poorly understood. Here, we show that an mRNA 3' processing factor, Fip1, is essential for embryonic stem cell (ESC) self-renewal and somatic cell reprogramming. Fip1 promotes stem cell maintenance, in part, by activating the ESC-specific APA profiles to ensure the optimal expression of a specific set of genes, including critical self-renewal factors. Fip1 expression and the Fip1-dependent APA program change during ESC differentiation and are restored to an ESC-like state during somatic reprogramming. Mechanistically, we provide evidence that the specificity of Fip1-mediated APA regulation depends on multiple factors, including Fip1-RNA interactions and the distance between APA sites. Together, our data highlight the role for post-transcriptional control in stem cell self-renewal, provide mechanistic insight on APA regulation in development, and establish an important function for APA in cell fate specification.

RNA-binding proteins in regulation of alternative cleavage and polyadenylation

Almost all eukaryotic pre-mRNAs are processed at the 3' end by the cleavage and polyadenylation (C/P) reaction, which preludes termination of transcription and gives rise to the poly(A) tail of mature mRNA. Genomic studies in recent years have indicated that most eukaryotic mRNA genes have multiple cleavage and polyadenylation sites (pAs), leading to alternative cleavage and polyadenylation (APA) products. APA isoforms generally differ in their 3' untranslated regions (3' UTRs), but can also have different coding sequences (CDSs). APA expands the repertoire of transcripts expressed from the genome, and is highly regulated under various physiological and pathological conditions. Growing lines of evidence have shown that RNA-binding proteins (RBPs) play important roles in regulation of APA. Some RBPs are part of the machinery for C/P; others influence pA choice through binding to adjacent regions. In this chapter, we review cis elements and trans factors involved in C/P, the significance of APA, and increasingly elucidated roles of RBPs in APA regulation. We also discuss analysis of APA using transcriptome-wide techniques as well as molecular biology approaches.

Genome-wide identification and predictive modeling of tissue-specific alternative polyadenylation

Motivation: Pre-mRNA cleavage and polyadenylation are essential steps for 3′-end maturation and subsequent stability and degradation of mRNAs. This process is highly controlled by cis-regulatory elements surrounding the cleavage/polyadenylation sites (polyA sites), which are frequently constrained by sequence content and position. More than 50% of human transcripts have multiple functional polyA sites, and the specific use of alternative polyA sites (APA) results in isoforms with variable 3′-untranslated regions, thus potentially affecting gene regulation. Elucidating the regulatory mechanisms underlying differential polyA preferences in multiple cell types has been hindered both by the lack of suitable data on the precise location of cleavage sites, as well as of appropriate tests for determining APAs with significant differences across multiple libraries.

Results: We applied a tailored paired-end RNA-seq protocol to specifically probe the position of polyA sites in three human adult tissue types. We specified a linear-effects regression model to identify tissue-specific biases indicating regulated APA; the significance of differences between tissue types was assessed by an appropriately designed permutation test. This combination allowed to identify highly specific subsets of APA events in the individual tissue types. Predictive models successfully classified constitutive polyA sites from a biologically relevant background (auROC = 99.6%), as well as tissue-specific regulated sets from each other. We found that the main cis-regulatory elements described for polyadenylation are a strong, and highly informative, hallmark for constitutive sites only. Tissue-specific regulated sites were found to contain other regulatory motifs, with the canonical polyadenylation signal being nearly absent at brain-specific polyA sites. Together, our results contribute to the understanding of the diversity of post-transcriptional gene regulation.

Availability: Raw data are deposited on SRA, accession numbers: brain SRX208132, kidney SRX208087 and liver SRX208134. Processed datasets as well as model code are published on our website: http://www.genome.duke.edu/labs/ohler/research/UTR/

Contact:uwe.ohler@duke.edu

Poly(A) motif prediction using spectral latent features from human DNA sequences

MOTIVATION

Polyadenylation is the addition of a poly(A) tail to an RNA molecule. Identifying DNA sequence motifs that signal the addition of poly(A) tails is essential to improved genome annotation and better understanding of the regulatory mechanisms and stability of mRNA. Existing poly(A) motif predictors demonstrate that information extracted from the surrounding nucleotide sequences of candidate poly(A) motifs can differentiate true motifs from the false ones to a great extent. A variety of sophisticated features has been explored, including sequential, structural, statistical, thermodynamic and evolutionary properties. However, most of these methods involve extensive manual feature engineering, which can be time-consuming and can require in-depth domain knowledge.

RESULTS

We propose a novel machine-learning method for poly(A) motif prediction by marrying generative learning (hidden Markov models) and discriminative learning (support vector machines). Generative learning provides a rich palette on which the uncertainty and diversity of sequence information can be handled, while discriminative learning allows the performance of the classification task to be directly optimized. Here, we used hidden Markov models for fitting the DNA sequence dynamics, and developed an efficient spectral algorithm for extracting latent variable information from these models. These spectral latent features were then fed into support vector machines to fine-tune the classification performance. We evaluated our proposed method on a comprehensive human poly(A) dataset that consists of 14 740 samples from 12 of the most abundant variants of human poly(A) motifs. Compared with one of the previous state-of-the-art methods in the literature (the random forest model with expert-crafted features), our method reduces the average error rate, false-negative rate and false-positive rate by 26, 15 and 35%, respectively. Meanwhile, our method makes ~30% fewer error predictions relative to the other string kernels. Furthermore, our method can be used to visualize the importance of oligomers and positions in predicting poly(A) motifs, from which we can observe a number of characteristics in the surrounding regions of true and false motifs that have not been reported before.

AVAILABILITY

http://sfb.kaust.edu.sa/Pages/Software.aspx.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Means to an end: mechanisms of alternative polyadenylation of messenger RNA precursors

Expression of mature messenger RNAs (mRNAs) requires appropriate transcription initiation and termination, as well as pre-mRNA processing by capping, splicing, cleavage, and polyadenylation. A core 3'-end processing complex carries out the cleavage and polyadenylation reactions, but many proteins have been implicated in the selection of polyadenylation sites among the multiple alternatives that eukaryotic genes typically have. In recent years, high-throughput approaches to map both the 3'-end processing sites as well as the binding sites of proteins that are involved in the selection of cleavage sites and in the processing reactions have been developed. Here, we review these approaches as well as the insights into the mechanisms of polyadenylation that emerged from genome-wide studies of polyadenylation across a range of cell types and states.

A unified model for yeast transcript definition

Identifying genes in the genomic context is central to a cell's ability to interpret the genome. Yet, in general, the signals used to define eukaryotic genes are poorly described. Here, we derived simple classifiers that identify where transcription will initiate and terminate using nucleic acid sequence features detectable by the yeast cell, which we integrate into a Unified Model (UM) that models transcription as a whole. The cis-elements that denote where transcription initiates function primarily through nucleosome depletion, and, using a synthetic promoter system, we show that most of these elements are sufficient to initiate transcription in vivo. Hrp1 binding sites are the major characteristic of terminators; these binding sites are often clustered in terminator regions and can terminate transcription bidirectionally. The UM predicts global transcript structure by modeling transcription of the genome using a hidden Markov model whose emissions are the outputs of the initiation and termination classifiers. We validated the novel predictions of the UM with available RNA-seq data and tested it further by directly comparing the transcript structure predicted by the model to the transcription generated by the cell for synthetic DNA segments of random design. We show that the UM identifies transcription start sites more accurately than the initiation classifier alone, indicating that the relative arrangement of promoter and terminator elements influences their function. Our model presents a concrete description of how the cell defines transcript units, explains the existence of nongenic transcripts, and provides insight into genome evolution.

A hybrid model for the prediction of mRNA polyadenylation signals

The mRNA polyadenylation is the cellular process that adds adenosine tails to mature mRNAs. Malfunction of polyadenylation has been implicated in several human diseases. In this paper, we proposed a novel feature extraction approach which employs the K-gram nucleotide acid pattern, the position weight matrix (PWM) and the increment of diversity (ID) to represent the original features. Then Principle Component Analysis (PCA) was applied to transform the original features into a new feature space where the low-dimensional features were used to train the real-coded genetic neural network model. In the experiments, our proposed algorithm (GA-BP) can achieve the accuracy about 82.98%, specificity 82.95% and sensitivity 83.01% in the specific dataset constructed by Kalkatawi. The results demonstrate that GA-BP is a promising algorithm for the prediction of mRNA polyadenylation signals.