TIGAR: transcript isoform abundance estimation method with gapped alignment of RNA-Seq data by variational Bayesian inference

BackgroundPhenotypic changes during cancer progression are associated with alterations in gene expression, which can be exploited to build molecular signatures for tumor stage identification and prognosis. However, it is not yet known whether the relative abundance of transcript isoforms may be informative for clinical stage and survival.MethodsUsing information theory and machine learning methods, we integrated RNA sequencing and clinical data from The Cancer Genome Atlas project to perform the first systematic analysis of the prognostic potential of transcript isoforms in 12 solid tumors to build new signatures for stage and prognosis. This study was also performed in breast tumors according to estrogen receptor (ER) status and melanoma tumors with proliferative and invasive phenotypes.ResultsTranscript isoform signatures accurately separate early from late-stage groups and metastatic from non-metastatic tumors, and are predictive of the survival of patients with undetermined lymph node invasion or metastatic status. These signatures show similar, and sometimes better, accuracies compared with known gene expression signatures in retrospective data and are largely independent of gene expression changes. Furthermore, we show frequent transcript isoform changes in breast tumors according to ER status, and in melanoma tumors according to the invasive or proliferative phenotype, and derive accurate predictive models of stage and survival within each patient subgroup.ConclusionsOur analyses reveal new signatures based on transcript isoform abundances that characterize tumor phenotypes and their progression independently of gene expression. Transcript isoform signatures appear especially relevant to determine lymph node invasion and metastasis and may potentially contribute towards current strategies of precision cancer medicine.Electronic supplementary materialThe online version of this article (doi:10.1186/s13073-016-0339-3) contains supplementary material, which is available to authorized users.

Compressive sensing is a sub-Nyquist sampling technique for efficient signal acquisition and reconstruction of sparse or compressible signals. In order to account for the sparsity of the underlying signal of interest, it is common to use sparsifying priors such as Bernoulli–Gaussian-inverse Gamma (BGiG) and Gaussian-inverse Gamma (GiG) priors on the components of the signal. With the introduction of variational Bayesian inference, the sparse Bayesian learning (SBL) methods for solving the inverse problem of compressive sensing have received significant interest as the SBL methods become more efficient in terms of execution time. In this paper, we consider the sparse signal recovery problem using compressive sensing and the variational Bayesian (VB) inference framework. More specifically, we consider two widely used Bayesian models of BGiG and GiG for modeling the underlying sparse signal for this problem. Although these two models have been widely used for sparse recovery problems under various signal structures, the question of which model can outperform the other for sparse signal recovery under no specific structure has yet to be fully addressed under the VB inference setting. Here, we study these two models specifically under VB inference in detail, provide some motivating examples regarding the issues in signal reconstruction that may occur under each model, perform comparisons and provide suggestions on how to improve the performance of each model.

Existing studies on sparse Bayesian learning for structural damage detection usually assume that the posterior probability density functions follow standard distributions which facilitate to circumvent the intractable integration problem of the evidence by means of numerical sampling or analytical derivation. Moreover, the uncertainties of each mode are usually quantified as a common parameter to simplify the calculation. These assumptions may not be realistic in practice. This study proposes a sparse Bayesian method for structural damage detection suitable for standard and nonstandard probability distributions. The uncertainty corresponding to each mode is assumed as different. Variational Bayesian inference is developed and the posterior probability density functions of each unknown are individually derived. The parameters are found to follow the gamma distribution, whereas the distribution of the damage index cannot be directly obtained because of the nonlinear relationship in its posterior probability density function. The delayed rejection adaptive Metropolis algorithm is then adopted to generate numerical samples of the damage index. The coupled damage index and parameters in the variational Bayesian inference are successively calculated via an iterative process. A laboratory-tested frame is utilised to verify the effectiveness of the proposed method. The results indicate that the sparse damage can be accurately detected. The proposed method has the advantage of high accuracy and broad applicability.

The fatty acid binding protein (FABP) family of proteins plays a role in multiple myeloma cell survival and disease progression in vitro and in silico, and thus blocking the FABPs could have anti-cancer effects and prove beneficial for myeloma patients.

Uncertainty propagation using infinite mixture of Gaussian processes and variational Bayesian inference

Accurate and comprehensive de novo transcriptome profiling in heart is a central issue to better understand cardiac physiology and diseases. Although significant progress has been made in genome-wide profiling for quantitative changes in cardiac gene expression, current knowledge offers limited insights to the total complexity in cardiac transcriptome at individual exon level. To develop more robust bioinformatic approaches to analyze high-throughput RNA sequencing (RNA-Seq) data, with the focus on the investigation of transcriptome complexity at individual exon and transcript levels. In addition to overall gene expression analysis, the methods developed in this study were used to analyze RNA-Seq data with respect to individual transcript isoforms, novel spliced exons, novel alternative terminal exons, novel transcript clusters (ie, novel genes), and long noncoding RNA genes. We applied these approaches to RNA-Seq data obtained from mouse hearts after pressure-overload-induced by transaortic constriction. Based on experimental validations, analyses of the features of the identified exons/transcripts, and expression analyses including previously published RNA-Seq data, we demonstrate that the methods are highly effective in detecting and quantifying individual exons and transcripts. Novel insights inferred from the examined aspects of the cardiac transcriptome open ways to further experimental investigations. Our work provided a comprehensive set of methods to analyze mouse cardiac transcriptome complexity at individual exon and transcript levels. Applications of the methods may infer important new insights to gene regulation in normal and disease hearts in terms of exon utilization and potential involvement of novel components of cardiac transcriptome.

Head and neck squamous cell carcinoma (HNSC) is a common malignancy with high mortality and poor prognosis. Alternative splicing (AS) is a transcriptional regulation mechanism that generates multiple transcripts from same genes, and aberrant AS signatures of cancers can be predictive for prognosis. We identified the survival-related AS events and splicing factors (SFs) from the RNA sequencing data and the corresponding clinical information of an HNSC cohort downloaded from The Cancer Genome Atlas (TCGA) and SpliceSeq. The independent prognostic predictors were assessed by Cox proportional regression analysis, and the regulatory network of SFs and AS events was analyzed by Spearman's test and constructed. A total of 4626 survival-related AS events in 3280 genes were identified, and most were protective factors. Among the different types of splicing events, exon skip was the most frequent. The prognostic models were constructed for each type of AS, and the area under the curve of the receiver operating characteristic curve of the combined prognostic model was 0.765, indicating good predictive performance. Finally, a correlation network between SF and AS events was constructed. We identified prognostic predictors based on AS events that stratified HNSC patients into the high- and low-risk groups, and revealed splicing networks that provide insights into the underlying mechanisms.

Since horse industry get bigger with horse racing and horse riding, athletic performance become most important trait on horses, but the molecular analysis and regulatory pathway studies remain insufficient for evaluation and prediction of horse athletic abilities. In our previous study, we conducted RNA-sequencing analysis with muscle and blood samples by exercise in Thoroughbreds. Through RNA-sequencing, we identified Pleckstrin Homology and RhoGEF Domain Containing G1 (PLEKHG1) gene expressed differentially by alternative spliced isoforms in skeletal muscle during exercise. In this study, we conducted reverse transcriptase polymerase chain reaction (RT-PCR) to validate two isoforms of equine PLEKHG1 transcripts (PLEKHG1a, PLEKHG1b), and cloned the transcripts to confirm the sequences. Additionally, we validated expression pattern of PLEKHG1a (long form of transcript) and PLEKHG1b (short form of transcript) in horse tissue by quantitative RT-PCR (qRT-PCR). Prediction of protein structure of these isoforms revealed two putative phosphorylation sites at the amino acid sequences encoded in exon 7, which is deleted in PLEKHG1b. Expression pattern of PLEKHG1a and PLEKHG1b shows cross expressed pattern as RNA-sequencing data. PLEKHG1a increased after exercise whereas PLEKHG1b decreased after exercise. Collectively, it is assumed that the expression patterns of PLEKHG1a and PLEKHG1b transcripts would be involved in regulation of myogenic differentiation and myogenic proliferation through the Ras signaling pathway. Further study should be necessary to uncover biological function(s) and significance of the alternative splicing isoforms in equine skeletal muscle.

In this letter, we propose a novel ship detection method in synthetic aperture radar (SAR) imagery via variational Bayesian inference. First, we establish the ship detection probabilistic model which decomposes the SAR image as the sum of a sparse component associated with ships and a sea clutter component. Then, we introduce hierarchical priors of the latent variables in the model and use variational Bayesian inference to estimate the posterior distributions of the latent variables. The proposed method is an automatic iterative process without any sliding window. Experimental results accomplished over synthetic data and a RADARSAT-2 SAR image demonstrate that the proposed method can achieve state-of-the-art ship detection performance.

This letter addresses the recovery of hierarchical sparse signals in a Bayesian framework. Hierarchical sparse signals exhibit two levels of sparsity, i.e., block-sparsity among different blocks and internal sparsity within each individual block. As in sparse Bayesian learning, each component of the coefficient vector is firstly modeled as a Gaussian distributed variable with zero mean. To enforce the two-level hierarchical sparsity, the variance is further modeled by two classes of hidden variables controlling the block-sparsity and the internal sparsity, respectively. Finally, variational Bayesian inference is used to recover the coefficient vector from the noise corrupted data. Numerical simulation and experimental results show that the proposed method outperforms those recently reported recovery methods.

We propose two generic methods for improving semi-supervised learning (SSL). The first integrates weight perturbation (WP) into existing “consistency regularization” (CR) based methods. We implement WP by leveraging variational Bayesian inference (VBI). The second method proposes a novel consistency loss called “maximum uncertainty regularization” (MUR). While most consistency losses act on perturbations in the vicinity of each data point, MUR actively searches for “virtual” points situated beyond this region that cause the most uncertain class predictions. This allows MUR to impose smoothness on a wider area in the input-output manifold. Our experiments show clear improvements in classification errors of various CR based methods when they are combined with VBI or MUR or both.

A low-complexity algorithm based on variational Bayesian inference for MIMO channel estimation

Cloud radio access network (C-RAN) is recognized as a key technology for the fifth generation (5G) wireless communication systems, where machine-type communication (MTC) is considered to support devices' connectivity. In this letter, we study the user activity detection (UAD) and channel estimation (CE) problems in C-RAN for MTC. Based on the user activity sparsity and signal spatial sparsity in C-RAN, we build a two-layer prior distribution graphical model to exploit the sparsity property and analyze the problem with variational Bayesian inference (VBI). We find that the width of prior distribution has a considerable impact on the algorithm performance and propose a modified VBI algorithm. Simulation results are presented to show that the proposed algorithm can achieve better performance with lower complexity than the existing approaches.

Identification of Recurrent Alternative RNA Splicing in Adverse-Risk Acute Myeloid Leukemia

Eukaryotic genes generate multiple RNA transcript isoforms though alternative transcription, splicing, and polyadenylation. However, the relationship between human transcript diversity and protein production is complex as each isoform can be translated differently. We fractionated a polysome profile and reconstructed transcript isoforms from each fraction, which we term Transcript Isoforms in Polysomes sequencing (TrIP-seq). Analysis of these data revealed regulatory features that control ribosome occupancy and translational output of each transcript isoform. We extracted a panel of 5' and 3' untranslated regions that control protein production from an unrelated gene in cells over a 100-fold range. Select 5' untranslated regions exert robust translational control between cell lines, while 3' untranslated regions can confer cell type-specific expression. These results expose the large dynamic range of transcript-isoform-specific translational control, identify isoform-specific sequences that control protein output in human cells, and demonstrate that transcript isoform diversity must be considered when relating RNA and protein levels.