Search

MotivationSequence-based protein–protein interaction (PPI) prediction represents a fundamental computational biology problem. To address this problem, extensive research efforts have been made to extract predefined features from the sequences. Based on these features, statistical algorithms are learned to classify the PPIs. However, such explicit features are usually costly to extract, and typically have limited coverage on the PPI information.ResultsWe present an end-to-end framework, PIPR (Protein–Protein Interaction Prediction Based on Siamese Residual RCNN), for PPI predictions using only the protein sequences. PIPR incorporates a deep residual recurrent convolutional neural network in the Siamese architecture, which leverages both robust local features and contextualized information, which are significant for capturing the mutual influence of proteins sequences. PIPR relieves the data pre-processing efforts that are required by other systems, and generalizes well to different application scenarios. Experimental evaluations show that PIPR outperforms various state-of-the-art systems on the binary PPI prediction problem. Moreover, it shows a promising performance on more challenging problems of interaction type prediction and binding affinity estimation, where existing approaches fall short.Availability and implementationThe implementation is available at https://github.com/muhaochen/seq_ppi.git.Supplementary informationSupplementary data are available at Bioinformatics online.

By implicitly recognizing a user based on his/her speech input, speaker identification enables many downstream applications, such as personalized system behavior and expedited shopping checkouts. Based on whether the speech content is constrained or not, both text-dependent (TD) and text-independent (TI) speaker recognition models may be used. We wish to combine the advantages of both types of models through an ensemble system to make more reliable predictions. However, any such combined approach has to be robust to incomplete inputs, i.e., when either TD or TI input is missing. As a solution we propose a fusion of embeddings network foenet architecture, combining joint learning with neural attention. We compare foenet with four competitive baseline methods on a dataset of voice assistant inputs, and show that it achieves higher accuracy than the baseline and score fusion methods, especially in the presence of incomplete inputs.

The widespread of Coronavirus has led to a worldwide pandemic with a high mortality rate. Currently, the knowledge accumulated from different studies about this virus is very limited. Leveraging a wide-range of biological knowledge, such as gene ontology and protein-protein interaction (PPI) networks from other closely related species presents a vital approach to infer the molecular impact of a new species. In this paper, we propose the transferred multi-relational embedding model Bio-JOIE to capture the knowledge of gene ontology and PPI networks, which demonstrates superb capability in modeling the SARS-CoV-2-human protein interactions. Bio-JOIE jointly trains two model components. The knowledge model encodes the relational facts from the protein and GO domains into separated embedding spaces, using a hierarchy-aware encoding technique employed for the GO terms. On top of that, the transfer model learns a non-linear transformation to transfer the knowledge of PPIs and gene ontology annotations across their embedding spaces. By leveraging only structured knowledge, Bio-JOIE significantly outperforms existing state-of-the-art methods in PPI type prediction on multiple species. Furthermore, we also demonstrate the potential of leveraging the learned representations on clustering proteins with enzymatic function into enzyme commission families. Finally, we show that Bio-JOIE can accurately identify PPIs between the SARS-CoV-2 proteins and human proteins, providing valuable insights for advancing research on this new disease., ACM BCB 2020, Best Student Paper

RNA-Sequencing has been the leading technology to quantify expression of thousands of genes simultaneously. The data analysis of an RNA-Seq experiment starts from aligning short reads to the reference genome/transcriptome or reconstructed transcriptome. However, current aligners lack the sensitivity to distinguish reads that come from homologous regions of an genome. One group of these homologies is the paralog pseudogenes. Pseudogenes arise from duplication of a set of protein coding genes, and have been considered as degraded paralogs in the genome due to their lost of functionality. Recent studies have provided evidence to support their novel regulatory roles in biological processes. With the growing interests in quantifying the expression level of pseudogenes at different tissues or cell lines, it is critical to have a sensitive method that can correctly align ambiguous reads and accurately estimate the expression level among homologous genes. Previously in PseudoLasso, we proposed a linear regression approach to learn read alignment behaviors, and to leverage this knowledge for abundance estimation and alignment correction. In this paper, we extend the work of PseudoLasso by grouping the homologous genomic regions into different communities using a community detection algorithm, followed by building a linear regression model separately for each community. The results show that this approach is able to retain the same accuracy as PseudoLasso. By breaking the genome into smaller homologous communities, the running time is improved from quadratic growth to linear with respect to the number of genes.

Motivation Circular RNA (circRNA) is a novel class of long non-coding RNAs that have been broadly discovered in the eukaryotic transcriptome. The circular structure arises from a non-canonical splicing process, where the donor site backspliced to an upstream acceptor site. These circRNA sequences are conserved across species. More importantly, rising evidence suggests their vital roles in gene regulation and association with diseases. As the fundamental effort toward elucidating their functions and mechanisms, several computational methods have been proposed to predict the circular structure from the primary sequence. Recently, advanced computational methods leverage deep learning to capture the relevant patterns from RNA sequences and model their interactions to facilitate the prediction. However, these methods fail to fully explore positional information of splice junctions and their deep interaction. Results We present a robust end-to-end framework, Junction Encoder with Deep Interaction (JEDI), for circRNA prediction using only nucleotide sequences. JEDI first leverages the attention mechanism to encode each junction site based on deep bidirectional recurrent neural networks and then presents the novel cross-attention layer to model deep interaction among these sites for backsplicing. Finally, JEDI can not only predict circRNAs but also interpret relationships among splice sites to discover backsplicing hotspots within a gene region. Experiments demonstrate JEDI significantly outperforms state-of-the-art approaches in circRNA prediction on both isoform level and gene level. Moreover, JEDI also shows promising results on zero-shot backsplicing discovery, where none of the existing approaches can achieve. Availability and implementation The implementation of our framework is available at https://github.com/hallogameboy/JEDI. Supplementary information Supplementary data are available at Bioinformatics online.

The functional impact of protein mutations is reflected on the alteration of conformation and thermodynamics of protein–protein interactions (PPIs). Quantifying the changes of two interacting proteins upon mutations is commonly carried out by computational approaches. Hence, extensive research efforts have been put to the extraction of energetic or structural features on proteins, followed by statistical learning methods to estimate the effects of mutations on PPI properties. Nonetheless, such features require extensive human labors and expert knowledge to obtain, and have limited abilities to reflect point mutations. We present an end-to-end deep learning framework, MuPIPR (Mutation Effects in Protein–protein Interaction PRediction Using Contextualized Representations), to estimate the effects of mutations on PPIs. MuPIPR incorporates a contextualized representation mechanism of amino acids to propagate the effects of a point mutation to surrounding amino acid representations, therefore amplifying the subtle change in a long protein sequence. On top of that, MuPIPR leverages a Siamese residual recurrent convolutional neural encoder to encode a wild-type protein pair and its mutation pair. Multi-layer perceptron regressors are applied to the protein pair representations to predict the quantifiable changes of PPI properties upon mutations. Experimental evaluations show that, with only sequence information, MuPIPR outperforms various state-of-the-art systems on estimating the changes of binding affinity for SKEMPI v1, and offers comparable performance on SKEMPI v2. Meanwhile, MuPIPR also demonstrates state-of-the-art performance on estimating the changes of buried surface areas. The software implementation is available at https://github.com/guangyu-zhou/MuPIPR.

Transposable elements (TE) comprise roughly half of the human genome. Though initially derided as junk DNA, they have been widely hypothesized to contribute to the evolution of gene regulation. However, the contribution of TE to the genetic architecture of diseases remains unknown. Here, we analyze data from 41 independent diseases and complex traits to draw three conclusions. First, TE are uniquely informative for disease heritability. Despite overall depletion for heritability (54% of SNPs, 39 ± 2% of heritability), TE explain substantially more heritability than expected based on their depletion for known functional annotations. This implies that TE acquire function in ways that differ from known functional annotations. Second, older TE contribute more to disease heritability, consistent with acquiring biological function. Third, Short Interspersed Nuclear Elements (SINE) are far more enriched for blood traits than for other traits. Our results can help elucidate the biological roles that TE play in the genetic architecture of diseases., Transposable elements (TE) make up a large component of the human genome and have been shown to contribute to human diseases. Here, Hormozdiari et al. estimate the contribution of TEs to the heritability of 41 complex traits and diseases and find enrichment of SINEs in blood traits.

Sequence-based protein-protein interaction (PPI) prediction represents a fundamental computational biology problem. To address this problem, extensive research efforts have been made to extract predefined features from the sequences. Based on these features, statistical algorithms are learned to classify the PPIs. However, such explicit features are usually costly to extract, and typically have limited coverage on the PPI information. Hence, we present an end-to-end framework, Lasagna, for PPI predictions using only the primary sequences of a protein pair. Lasagna incorporates a deep residual recurrent convolutional neural network in the Siamese learning architecture, which leverages both robust local features and contextualized information that are significant for capturing the mutual influence of protein sequences. Our framework relieves the data pre-processing efforts that are required by other systems, and generalizes well to different application scenarios. Experimental evaluations show that Lasagna outperforms various state-of-the-art systems on the binary PPI prediction problem. Moreover, it shows a promising performance on more challenging problems of interaction type prediction and binding affinity estimation, where existing approaches fall short.

k-mer profiling has been one of the trending approaches to analyze read data generated by high-throughput sequencing technologies. The tasks of k-mer profiling include, but are not limited to, counting the frequencies and determining the occurrences of short sequences in a dataset. The notion of k-mer has been extensively used to build de Bruijn graphs in genome or transcriptome assembly, which requires examining all possible k-mers presented in the dataset. Recently, an alternative way of profiling has been proposed, which constructs a set of representative k-mers as genomic markers and profiles their occurrences in the sequencing data. This technique has been applied in both transcript quantification through RNA-Seq and taxonomic classification of metagenomic reads. Most of these applications use a set of fixed-size k-mers since the majority of existing k-mer counters are inadequate to process genomic sequences with variable-length k-mers. However, choosing the appropriate k is challenging, as it varies for different applications. As a pioneer work to profile a set of variable-length k-mers, we propose TahcoRoll in order to enhance the Aho-Corasick algorithm. More specifically, we use one bit to represent each nucleotide, and integrate the rolling hash technique to construct an efficient in-memory data structure for this task. Using both synthetic and real datasets, results show that TahcoRoll outperforms existing approaches in either or both time and memory efficiency without using any disk space. In addition, compared to the most efficient state-of-the-art k-mer counters, such as KMC and MSBWT, TahcoRoll is the only approach that can process long read data from both PacBio and Oxford Nanopore on a commodity desktop computer. The source code of TahcoRoll is implemented in C++14, and available at https://github.com/chelseaju/TahcoRoll.git.

There is increasing evidence that many risk loci found using genome-wide association studies are molecular quantitative trait loci (QTLs). Here we introduce a new set of functional annotations based on causal posterior probabilities of fine-mapped molecular cis-QTLs, using data from the Genotype-Tissue Expression (GTEx) and BLUEPRINT consortia. We show that these annotations are more strongly enriched for heritability (5.84× for eQTLs; P = 1.19 × 10-31) across 41 diseases and complex traits than annotations containing all significant molecular QTLs (1.80× for expression (e)QTLs). eQTL annotations obtained by meta-analyzing all GTEx tissues generally performed best, whereas tissue-specific eQTL annotations produced stronger enrichments for blood- and brain-related diseases and traits. eQTL annotations restricted to loss-of-function intolerant genes were even more enriched for heritability (17.06×; P = 1.20 × 10-35). All molecular QTLs except splicing QTLs remained significantly enriched in joint analysis, indicating that each of these annotations is uniquely informative for disease and complex trait architectures.

Recent successes in genome-wide association studies (GWASs) make it possible to address important questions about the genetic architecture of complex traits, such as allele frequency and effect size. One lesser-known aspect of complex traits is the extent of allelic heterogeneity (AH) arising from multiple causal variants at a locus. We developed a computational method to infer the probability of AH and applied it to three GWAS and four expression quantitative trait loci (eQTL) datasets. We identified a total of 4152 loci with strong evidence of AH. The proportion of all loci with identified AH is 4-23% in eQTLs, 35% in GWAS of High-Density Lipoprotein (HDL), and 23% in schizophrenia. For eQTLs, we observed a strong correlation between sample size and the proportion of loci with AH (R2=0.85, P = 2.2e-16), indicating that statistical power prevents identification of AH in other loci. Understanding the extent of AH may guide the development of new methods for fine mapping and association mapping of complex traits.

Bud formation is an adaptive trait that temperate forest trees have acquired to facilitate seasonal synchronization. We have characterized transcriptome-level changes that occur during bud formation of white spruce [Picea glauca (Moench) Voss], a primarily determinate species in which preformed stem units contained within the apical bud constitute most of next season's growth. Microarray analysis identified 4460 differentially expressed sequences in shoot tips during short day-induced bud formation. Cluster analysis revealed distinct temporal patterns of expression, and functional classification of genes in these clusters implied molecular processes that coincide with anatomical changes occurring in the developing bud. Comparing expression profiles in developing buds under long day and short day conditions identified possible photoperiod-responsive genes that may not be essential for bud development. Several genes putatively associated with hormone signalling were identified, and hormone quantification revealed distinct profiles for abscisic acid (ABA), cytokinins, auxin and their metabolites that can be related to morphological changes to the bud. Comparison of gene expression profiles during bud formation in different tissues revealed 108 genes that are differentially expressed only in developing buds and show greater transcript abundance in developing buds than other tissues. These findings provide a temporal roadmap of bud formation in white spruce.

Metabolomics investigations hold promise for the characterization of small molecules, metabolites, which govern the ultimate manifestation of cardiac phenotypes. In this study, we employed a mass spectrometry-based metabolomics approach to identify metabolic marker(s), which dynamically reflect the cardiac performance of heart failure patients amid the implantation of mechanical circulatory support. Using the MRM-based and triple quadrupole technology platform, we have quantified 266 metabolites native to human plasma and collected from thirteen heart failure patients. The temporal profile of these metabolites was sampled from 1 day prior to the implantation of mechanical circulatory support, as well as 1-, 3-, 5-, and 7-day following their surgical interventions. We identified subgroups of these metabolites with coordinated behaviors that are interesting to their diseased phenotypes. In a pair-wise correlation analysis, 36.8% (98 out of 266) of metabolites were significantly correlated. Intriguingly, majority of which (65 out of 98) are representing the functional groups of phosphatidylcholines; several of them are known to have close associations with the pathogenesis of cardiovascular diseases. In addition, there are 33 metabolites contributing to multiple functional groups, including twelve of them belong to sphingomyelines, ten of them in the family of lysophosphatidylcholines, eight amino acids (Gln, Ser, Ala, His, Lys, Gly, Thr, and Arg), as well as three fatty acids (eicosapentaenoic acid, pentadecenoic acid, and heptadecenoic acid). The behaviors of these 266 metabolites have constituted individualized metabolic fingerprints. Delineation of the intrinsic relationships among alterations in distinct metabolite groups and their reflected cardiac function will enable us to identify new metabolic markers aiding stratification and/or prediction on the clinical outcome of each individual patient undergoing the treatment of mechanical circulatory support. This personalized metabolic fingerprint will offer unique prognostic utilities, supporting clinical decision-making process to deliver intervention that is most effective and beneficial to an individual.