Biology subjects

Li, J. J.

Publications and source records attributed to Li, J. J..

67 records · Page 4Linked to original sources

bioRxiv2020

Annotating Gene Ontology terms for protein sequences with the Transformer model

Predicting functions for novel amino acid sequences is a long-standing research problem. The Uniprot database which contains protein sequences annotated with Gene Ontology (GO) terms, is one commonly used training dataset for this problem. Predicting protein functions can then be viewed as a multi-label classification problem where the input is an amino acid sequence and the output is a set of GO terms. Recently, deep convolutional neural network (CNN) models have been introduced to annotate GO terms for protein sequences. However, the CNN architecture can only model close-range interactions between amino acids in a sequence. In this paper, first, we build a novel GO annotation model based on the Transformer neural network. Unlike the CNN architecture, the Transformer models all pairwise interactions for the amino acids within a sequence, and so can capture more relevant information from the sequences. Indeed, we show that our adaptation of Transformer yields higher classification accuracy when compared to the recent CNN-based method DeepGO. Second, we modify our model to take motifs in the protein sequences found by BLAST as additional input features. Our strategy is different from other ensemble approaches that average the outcomes of BLAST-based and machine learning predictors. Third, we integrate into our Transformer the metadata about the protein sequences such as 3D structure and protein-protein interaction (PPI) data. We show that such information can greatly improve the prediction accuracy, especially for rare GO labels.

bioinformatics↗

bioRxiv2019

Bipartite Tight Spectral Clustering (BiTSC) Algorithm for Identifying Conserved Gene Co-clusters in Two Species

Gene clustering is a widely-used technique that has enabled computational prediction of unknown gene functions within a species. However, it remains a challenge to refine gene function prediction by leveraging evolutionarily conserved genes in another species. This challenge calls for a new computational algorithm to identify gene co-clusters in two species, so that genes in each co-cluster exhibit similar expression levels in each species and strong conservation between the species. Here we develop the bipartite tight spectral clustering (BiTSC) algorithm, which identifies gene co-clusters in two species based on gene orthology information and gene expression data. BiTSC novelly implements a formulation that encodes gene orthology as a bipartite network and gene expression data as node covariates. This formulation allows BiTSC to adopt and combine the advantages of multiple unsupervised learning techniques: kernel enhancement, bipartite spectral clustering, consensus clustering, tight clustering, and hierarchical clustering. As a result, BiTSC is a flexible and robust algorithm capable of identifying informative gene co-clusters without forcing all genes into co-clusters. Another advantage of BiTSC is that it does not rely on any distributional assumptions. Beyond cross-species gene co-clustering, BiTSC also has wide applications as a general algorithm for identifying tight node co-clusters in any bipartite network with node covariates. We demonstrate the accuracy and robustness of BiTSC through comprehensive simulation studies. In a real data example, we use BiTSC to identify conserved gene co-clusters of D. melanogaster and C. elegans, and we perform a series of downstream analysis to both validate BiTSC and verify the biological significance of the identified co-clusters.

bioinformatics↗

bioRxiv2019

Detecting Local Genetic Correlations with Scan Statistics

Genetic correlation analysis has quickly gained popularity in the past few years and provided insights into the genetic etiology of numerous complex diseases. However, existing approaches oversimplify the shared genetic architecture between different phenotypes and cannot effectively identify precise genetic regions contributing to the genetic correlation. In this work, we introduce LOGODetect, a powerful and efficient statistical method to identify small genome segments harboring local genetic correlation signals. LOGODetect automatically identifies genetic regions showing consistent associations with multiple phenotypes through a scan statistic approach. It uses summary association statistics from genome-wide association studies (GWAS) as input and is robust to sample overlap between studies. Applied to five phenotypically distinct but genetically correlated psychiatric disorders, we identified 49 non-overlapping genome regions associated with multiple disorders, including multiple hub regions showing concordant effects on more than two disorders. Our method addresses critical limitations in existing analytic strategies and may have wide applications in post-GWAS analysis.

genetics↗

bioRxiv2019

Learning Representations for Gene Ontology Terms by Contextualized Text Encoders

Protein functions can be described by the Gene Ontology (GO) terms, allowing us to compare the functions of two proteins by measuring the similarity of the terms assigned to them. Recent works have applied neural network models to derive the vector representations for GO terms and compute similarity scores for these terms by comparing their vector embeddings. There are two typical ways to embed GO terms into vectors; a model can either embed the definitions of the terms or the topology of the terms in the ontology. In this paper, we design three tasks to critically evaluate the GO embeddings of two recent neural network models, and further introduce additional models for embedding GO terms, adapted from three popular neural network frameworks: Graph Convolution Network (GCN), Embeddings from Language Models (ELMo), and Bidirectional Encoder Representations from Transformers (BERT), which have not yet been explored in previous works. Task 1 studies edge cases where the GO embeddings may not provide meaningful similarity scores for GO terms. We find that all neural network based methods fail to produce high similarity scores for related terms when these terms have low Information Content values. Task 2 is a canonical task which estimates how well GO embeddings can compare functions of two orthologous genes or two interacting proteins. The best neural network methods for this task are those that embed GO terms using their definitions, and the differences among such methods are small. Task 3 evaluates how GO embeddings affect the performance of GO annotation methods, which predict whether a protein should be labeled by certain GO terms. When the annotation datasets contain many samples for each GO label, GO embeddings do not improve the classification accuracy. Machine learning GO annotation methods often remove rare GO labels from the training datasets so that the model parameters can be efficiently trained. We evaluate whether GO embeddings can improve prediction of rare labels unseen in the training datasets, and find that GO embeddings based on the BERT framework achieve the best results in this setting. We present our embedding methods and three evaluation tasks as the basis for future research on this topic.

bioinformatics↗

bioRxiv2019

The Positive End of the Polygenic Score Distribution for ADHD: A Risk, Protective or Resilience Factor?

BackgroundPolygenic scores (PGS) are widely used to characterize genetic liability for heritable mental disorders, including attention-deficit/hyperactivity disorder (ADHD). However, little is known about the effects of a low burden of genetic liability for ADHD, including whether this functions as a low risk or protective factor for ADHD and related functional outcomes in later life. The current study examines the association of low ADHD PGS and functional outcomes in adulthood.\n\nMethodsParticipants were from Wave IV of the National Longitudinal Study of Adolescent to Adult Health (Add Health) (N=7,088; mean age=29, s.d.=1.74). ADHD PGS was computed from an existing genome-wide association study, and adult functional outcomes, including cognition, educational attainment, mental health and physical health were assessed during in-home interviews.\n\nResultsIndividuals at the lowest end of the ADHD PGS distribution (i.e., lowest 20th percentile) had the lowest probabilities of ADHD, exhibiting a 17-19% reduction in risk for ADHD relative to the observed 8.3% prevalence rate of ADHD in Add Health. Furthermore, individuals with low ADHD PGS had higher cognitive performance, greater levels of educational attainment, and lower BMI relative to individuals representing the rest of the ADHD PGS distribution, including those who were in the medium and high PGS groups.\n\nConclusionsFindings indicate that psychiatric PGS likely capture far more than just the risk and the absence of risk for a psychiatric outcome; where one lies along the PGS distribution may predict diverging functional consequences, for better and for worse.

genomics↗

bioRxiv2019

Quantitation of cis-translational control by general mRNA sequence features in five eukaryotes

BACKGROUNDGeneral translational cis-elements are present in the mRNAs of all genes and affect the recruitment, assembly, and progress of preinitiation complexes and the ribosome under many physiological states. These elements are: mRNA folding, upstream open reading frames, specific nucleotides flanking the initiating AUG codon, protein coding sequence length, and codon usage. The quantitative contributions of these sequence features and how and why they coordinate together to control translation rates are not well understood.\n\nRESULTSHere we show that these sequence features specify 42%-81% of the variance in translation rates in S. cerevisiae, S. pombe, Arabidopsis thaliana, M. musculus, and H. Sapiens. We establish that control by RNA secondary structure is chiefly mediated by highly folded 25-60 nucleotide segments within mRNA 5 regions; that changes in tri-nucleotide frequencies between highly and poorly translated 5 regions are correlated between all species; and that control by distinct biochemical processes is extensively correlated as is regulation by a single process acting in different parts of the same mRNA.\n\nCONCLUSIONSOur work shows that the general features control a much larger fraction of the variance in translation rates than previously realized. We provide a more detailed and accurate understanding of the aspects of RNA structure that direct translation in diverse eukaryotes. In addition, we note that the strongly correlated regulation between and within cis-control features will cause more even densities of translational complexes along each mRNA and therefore more efficient use of the translation machinery by the cell.

genomics↗

bioRxiv2019

EpiAlign: an alignment-based bioinformatic tool for comparing chromatin state sequences

The availability of genome-wide epigenomic datasets enables in-depth studies of epigenetic modifications and their relationships with chromatin structures and gene expression. Various alignment tools have been developed to align nucleotide or protein sequences in order to identify structurally similar regions. However, there are currently no alignment methods specifically designed for comparing multi-track epigenomic signals and detecting common patterns that may explain functional or evolutionary similarities. We propose a new local alignment algorithm, EpiAlign, designed to compare chromatin state sequences learned from multi-track epigenomic signals and to identify locally aligned chromatin regions. EpiAlign is a dynamic programming algorithm that novelly incorporates varying lengths and frequencies of chromatin states. We demonstrate the effcacy of EpiAlign through extensive simulations and studies on the real data from the NIH Roadmap Epigenomics project. EpiAlign is able to extract recurrent chromatin state patterns along a single epigenome, and many of these patterns carry cell-type-specific characteristics. EpiAlign can also detect common chromatin state patterns across multiple epigenomes, and it will serve as a useful tool to group and distinguish epigenomic samples based on genome-wide or local chromatin state patterns.

bioinformatics↗

bioRxiv2018

A statistical simulator scDesign for rational scRNA-seq experimental design

Single-cell RNA-sequencing (scRNA-seq) has revolutionized biological sciences by revealing genome-wide gene expression levels within an individual cell. However, a critical challenge faced by researchers is how to optimize the choices of sequencing platforms, sequencing depths, and cell numbers in designing scRNA-seq experiments, so as to balance the exploration of the depth and breadth of transcriptome information. Here we present a flexible and robust simulator, scDesign, the first statistical framework for researchers to quantitatively assess practical scRNA-seq experimental design in the context of differential gene expression analysis. In addition to experimental design, scDesign also assists computational method development by generating high-quality synthetic scRNA-seq datasets under customized experimental settings. In an evaluation based on 17 cell types and six different protocols, scDesign outperformed four state-of-the-art scRNA-seq simulation methods and led to rational experimental design.

bioinformatics↗

bioRxiv2018

AIDE: annotation-assisted isoform discovery and abundanceestimation from RNA-seq data

Genome-wide accurate identification and quantification of full-length mRNA isoforms is crucial for investigating transcriptional and post-transcriptional regulatory mechanisms of biological phenomena. Despite continuing efforts in developing effective computational tools to identify or assemble full-length mRNA isoforms from second-generation RNA-seq data, it remains a challenge to accurately identify mRNA isoforms from short sequence reads due to the substantial information loss in RNA-seq experiments. Here we introduce a novel statistical method, AIDE (Annotation-assisted Isoform DiscovEry), the first approach that directly controls false isoform discoveries by implementing the testing-based model selection principle. Solving the isoform discovery problem in a stepwise and conservative manner, AIDE prioritizes the annotated isoforms and precisely identifies novel isoforms whose addition significantly improves the explanation of observed RNA-seq reads. We evaluate the performance of AIDE based on multiple simulated and real RNA-seq datasets followed by a PCR-Sanger sequencing validation. Our results show that AIDE effectively leverages the annotation information to compensate the information loss due to short read lengths. AIDE achieves the highest precision in isoform discovery and the lowest error rates in isoform abundance estimation, compared with three state-of-the-art methods Cufflinks, SLIDE, and StringTie. As a robust bioinformatics tool for transcriptome analysis, AIDE will enable researchers to discover novel transcripts with high confidence.

bioinformatics↗

bioRxiv2018

Genome-wide association study reveals sex-specific genetic architecture of facial attractiveness

Facial attractiveness is a complex human trait of great interest in both academia and industry. Literature on sociological and phenotypic factors associated with facial attractiveness is rich, but its genetic basis is poorly understood. In this paper, we conducted a genome-wide association study to discover genetic variants associated with facial attractiveness using 3,928 samples in the Wisconsin Longitudinal Study. We identified two genome-wide significant loci and highlighted a handful of candidate genes, many of which are specifically expressed in human tissues involved in reproduction and hormone synthesis. Additionally, facial attractiveness showed strong and negative genetic correlations with BMI in females and with blood lipids in males. Our analysis also suggested sex-specific selection pressure on variants associated with lower male attractiveness. These results revealed sex-specific genetic architecture of facial attractiveness and provided fundamental new insights into its genetic basis.

genetics↗

bioRxiv2017

Spliceosome profiling visualizes the operations of a dynamic RNP in vivo at nucleotide resolution

Tools to understand how the spliceosome functions in vivo have lagged behind advances in its structural biology. We describe methods to globally profile spliceosome-bound precursor, intermediates and products at nucleotide resolution. We apply these tools to three divergent yeast species that span 600 million years of evolution. The sensitivity of the approach enables detection of novel cases of non-canonical catalysis including interrupted, recursive and nested splicing. Employing statistical modeling to understand the quantitative relationships between RNA features and the data, we uncover independent roles for intron size, position and number in substrate progression through the two catalytic stages. These include species-specific inputs suggestive of spliceosome-transcriptome coevolution. Further investigations reveal ATP-dependent discard of numerous endogenous substrates at both the precursor and lariat-intermediate stages and connect discard to intron retention, a form of splicing regulation. Spliceosome profiling is a quantitative, generalizable global technology to investigate an RNP central to eukaryotic gene expression.\n\nHighlightsO_LIMeasurement of spliceosome-bound precursor and intermediate in three species\nC_LIO_LINon-canonical splicing events revealed\nC_LIO_LIStatistical modeling uncovers substrate features that predict catalytic efficiency\nC_LIO_LIDiscard of suboptimal substrates occurs in vivo and predicts intron-retained mRNAs\nC_LI

molecular biology↗

bioRxiv2017

scImpute: Accurate And Robust Imputation For Single Cell RNA-Seq Data

The emerging single cell RNA sequencing (scRNA-seq) technologies enable the investigation of transcriptomic landscapes at single-cell resolution. The analysis of scRNA-seq data is complicated by excess zero or near zero counts, the so-called dropouts due to the low amounts of mRNA sequenced within individual cells. Downstream analysis of scRNA-seq would be severely biased if the dropout events are not properly corrected. We introduce scImpute, a statistical method to accurately and robustly impute the dropout values in scRNA-seq data. ScImpute automatically identifies gene expression values affected by dropout events, and only perform imputation on these values without introducing new bias to the rest data. ScImpute also detects outlier or rare cells and excludes them from imputation. Evaluation based on both simulated and real scRNA-seq data on mouse embryos, mouse brain cells, human blood cells, and human embryonic stem cells suggests that scImpute is an effective tool to recover transcriptome dynamics masked by dropout events. scImpute is shown to correct false zero counts, enhance the clustering of cell populations and subpopulations, improve the accuracy of differential expression analysis, and aid the study of gene expression dynamics.

bioinformatics↗

bioRxiv2017

Quantitating Translational Control: mRNA Abundance-Dependent and Independent Contributions

Translation rate per mRNA molecule correlates positively with mRNA abundance. As a result, protein levels do not scale linearly with mRNA levels, but instead scale with the abundance of mRNA raised to the power of an \"amplification exponent\". Here we show that to quantitate translational control, the translation rate must be decomposed into two components. One, TRmD, depends on the mRNA level and defines the amplification exponent. The other, TRmIND, is independent of mRNA amount and impacts the correlation coefficient between protein and mRNA levels. We show that in S. cerevisiae TRmD represents [~]20% of the variance in translation and directs an amplification exponent of 1.20 with a 95% confidence interval [1.14, 1.26]. TRmIND constitutes the remaining [~]80% of the variance in translation and explains [~]5% of the variance in protein expression. We also find that TRmD and TRmIND are preferentially determined by different mRNA sequence features: TRmIND by the length of the open reading frame and TRmD both by a [~]60 nucleotide element that spans the initiating AUG and by codon and amino acid frequency. Our work provides more appropriate estimates of translational control and implies that TRmIND is under different evolutionary selective pressures than TRmD.

genomics↗