Journal of Bioinformatics and Computational Biology

Most of the current computational models for splice junction prediction are based on the identification of canonical splice junctions. However, it is observed that the junctions lacking the consensus dimers GT and AG also undergo splicing. Identification of such splice junctions, called the non-canonical splice junctions, is also essential for a comprehensive understanding of the splicing phenomenon. This work focuses on the identification of non-canonical splice junctions through the application of a bidirectional long short-term memory (BLSTM) network. Furthermore, we apply a back-propagation-based (integrated gradient) and a perturbation-based (occlusion) visualization techniques to extract the non-canonical splicing features learned by the model. The features obtained are validated with the existing knowledge from the literature. Integrated gradient extracts features that comprise contiguous nucleotides, whereas occlusion extracts features that are individual nucleotides distributed across the sequence.

COVID-19 pandemic has caused a global health crisis. Developing vaccines would need a good knowledge of genetic properties of SARS-CoV-2. The most fundamental approach is to look into the structures of its RNA, in particular, the nucleotides and amino acids. This motivates our research on this topic. We study the occurrence structures of nitrogenous bases and amino acids. To this aim, we devise a structural metric which could measure the structure differences for bases or amino acids. By analyzing various SARS-CoV-2 samples, we calculate the distance matrices for nitrogenous bases and amino acids. Based on the distance matrices, we find the average distance matrices for them, respectively. Then we identify the relations of all the minimal distances between bases and amino acids. The results also show that different substructures would yield much more diversified distances between amino acids. In the end, we also conduct the comparison of our structural metric with other frequently used metrics, in particular, Hausdorff metrics.

Viral infection is a complicated dynamic process, in which viruses intrude into cells to duplicate themselves and trigger succeeding biological processes regulated by genes. It may lead to a serious disaster to human’s health. A scheme is proposed to monitor the response of cells after being infected by viruses. Co-expression levels of genes measured at successive time points form a gene expression profile sequence, which is mapped to a temporal gene regulatory network. The fission and fusion of the communities of the networks are used to find the active parts. We investigated an experiment of injection of flu viruses into a total of 17 healthy volunteers, which develop into an infected group and a survival group. The survival group is much more chaotic, i.e. there occur complicated fissions and fusions of communities over the whole network. For the infected group, the most active part of the regulatory network forms a single community, but it is included in one of the large communities and completely conservative in the survival group. There are a total of six and seven genes in the active structure that take part in the Parkinson’s disease and the ribosome pathways, respectively. Actually, a total of 30 genes (covering $30/48=62.5\%$) of the genes in the active structure participate in the neuro-degeneration and its related pathways. This scheme can be extended straightforwardly to extract characteristics of trajectories of complex systems.

Neuroimaging genetics has become an important research topic since it can reveal complex associations between genetic variants (i.e. single nucleotide polymorphisms (SNPs) and the structures or functions of the human brain. However, existing kernel mapping is difficult to directly use the sparse representation method in the kernel feature space, which makes it difficult for most existing sparse canonical correlation analysis (SCCA) methods to be directly promoted in the kernel feature space. To bridge this gap, we adopt a novel alternating projected gradient approach, gradient KCCA (gradKCCA) model to develop a powerful model for exploring the intrinsic associations among genetic markers, imaging quantitative traits (QTs) of interest. Specifically, this model solves kernel canonical correlation (KCCA) with an additional constraint that projection directions have pre-images in the original data space, a sparsity-inducing variant of the model is achieved through controlling the ${\ell }_1$-norm of the preimages of the projection directions. We evaluate this model using Alzheimer’s disease Neuroimaging Initiative (ADNI) cohort to discover the relationships among SNPs from Alzheimer’s disease (AD) risk gene APOE, imaging QTs extracted from structural magnetic resonance imaging (MRI) scans. Our results show that the algorithm not only outperforms the traditional KCCA method in terms of Root Mean Square Error (RMSE) and Correlation Coefficient (CC) but also identify the meaningful and relevant biomarkers of SNPs (e.g. rs157594 and rs405697), which are positively related to right Postcentral and right SupraMarginal brain regions in this study. Empirical results indicate its promising capability in revealing biologically meaningful neuroimaging genetics associations and improving the disease-related mechanistic understanding of AD.

Prostate Specific Antigen (PSA) level in the serum is one of the most widely used markers in monitoring prostate cancer (PCa) progression, treatment response, and disease relapse. Although significant efforts have been taken to analyze various socioeconomic and cultural factors that contribute to the racial disparities in PCa, limited research has been performed to quantitatively understand how and to what extent molecular alterations may impact differential PSA levels present at varied tumor status between African–American and European–American men. Moreover, missing values among patients add another layer of difficulty in precisely inferring their outcomes. In light of these issues, we propose a data-driven, deep learning-based imputation and inference framework (DIIF). DIIF seamlessly encapsulates two modules: an imputation module driven by a regularized deep autoencoder for imputing critical missing information and an inference module in which two deep variational autoencoders are coupled with a graphical inference model to quantify the personalized and race-specific causal effects. Large-scale empirical studies on the independent sub-cohorts of The Cancer Genome Atlas (TCGA) PCa patients demonstrate the effectiveness of DIIF. We further found that somatic mutations in TP53, ATM, PTEN, FOXA1, and PIK3CA are statistically significant genomic factors that may explain the racial disparities in different PCa features characterized by PSA.

Accurately determining a change in protein binding affinity upon mutations is important to find novel therapeutics and to assist mutagenesis studies. Determination of change in binding affinity upon mutations requires sophisticated, expensive, and time-consuming wet-lab experiments that can be supported with computational methods. Most of the available computational prediction techniques depend upon protein structures that bound their applicability to only protein complexes with recognized 3D structures. In this work, we explore the sequence-based prediction of change in protein binding affinity upon mutation and question the effectiveness of $K$-fold cross-validation (CV) across mutations adopted in previous studies to assess the generalization ability of such predictors with no known mutation during training. We have used protein sequence information instead of protein structures along with machine learning techniques to accurately predict the change in protein binding affinity upon mutation. Our proposed sequence-based novel change in protein binding affinity predictor called PANDA performs comparably to the existing methods gauged through an appropriate CV scheme and an external independent test dataset. On an external test dataset, our proposed method gives a maximum Pearson correlation coefficient of 0.52 in comparison to the state-of-the-art existing protein structure-based method called MutaBind which gives a maximum Pearson correlation coefficient of 0.59. Our proposed protein sequence-based method, to predict a change in binding affinity upon mutations, has wide applicability and comparable performance in comparison to existing protein structure-based methods. We made PANDA easily accessible through a cloud-based webserver and python code available at https://sites.google.com/view/wajidarshad/software and https://github.com/wajidarshad/panda, respectively.

In the field of comparative genomics, one way of comparing two genomes is through the analysis of how they distinguish themselves based on a set of mutations called rearrangement events. When considering that genomes undergo different types of rearrangements, it can be assumed that some events are more common than others. To model this assumption, one can assign different weights to different events, where common events tend to cost less than others. However, this approach, called weighted, does not guarantee that the rearrangement assumed to be the most frequent will be also the most frequently returned by proposed algorithms. To overcome this issue, we investigate a new problem where we seek the shortest sequence of rearrangement events able to transform one genome into the other, with a restriction regarding the proportion between the events returned. Here, we consider two rearrangement events: reversal, that inverts the order and the orientation of the genes inside a segment of the genome, and transposition, that moves a segment of the genome to another position. We show the complexity of this problem for any desired proportion considering scenarios where the orientation of the genes is known or unknown. We also develop an approximation algorithm with a constant approximation factor for each scenario and, in particular, we describe an improved (asymptotic) approximation algorithm for the case where the gene orientation is known. At last, we present the experimental tests comparing the proposed algorithms with others from the literature for the version of the problem without the proportion restriction.

DNA-binding proteins (DBPs) perform an influential role in diverse biological activities like DNA replication, slicing, repair, and transcription. Some DBPs are indispensable for understanding many types of human cancers (i.e. lung, breast, and liver cancer) and chronic diseases (i.e. AIDS/HIV, asthma), while other kinds are involved in antibiotics, steroids, and anti-inflammatory drugs designing. These crucial processes are closely related to DBPs types. DBPs are categorized into single-stranded DNA-binding proteins (ssDBPs) and double-stranded DNA-binding proteins (dsDBPs). Few computational predictors have been reported for discriminating ssDBPs and dsDBPs. However, due to the limitations of the existing methods, an intelligent computational system is still highly desirable. In this work, features from protein sequences are discovered by extending the notion of dipeptide composition (DPC), evolutionary difference formula (EDF), and K-separated bigram (KSB) into the position-specific scoring matrix (PSSM). The highly intrinsic information was encoded by a compression approach named discrete cosine transform (DCT) and the model was trained with support vector machine (SVM). The prediction performance was further boosted by the genetic algorithm (GA) ensemble strategy. The novel predictor (DBP-GAPred) acquired 1.89%, 0.28%, and 6.63% higher accuracies on jackknife, 10-fold, and independent dataset tests, respectively than the best predictor. These outcomes confirm the superiority of our method over the existing predictors.