Genome-Wide Functional Annotation of Human Protein-Coding Splice Variants Using Multiple Instance Learning

Abstract

The vast majority of human multiexon genes undergo alternative splicing and produce a variety of splice variant transcripts and proteins, which can perform different functions. These protein-coding splice variants (PCSVs) greatly increase the functional diversity of proteins. Most functional annotation algorithms have been developed at the gene level; the lack of isoform-level gold standards is an important intellectual limitation for currently available machine learning algorithms. The accumulation of a large amount of RNA-seq data in the public domain greatly increases our ability to examine the functional annotation of genes at isoform level. In the present study, we used a multiple instance learning (MIL)-based approach for predicting the function of PCSVs. We used transcript-level expression values and gene-level functional associations from the Gene Ontology database. A support vector machine (SVM)-based 5-fold cross-validation technique was applied. Comparatively, genes with multiple PCSVs performed better than single PCSV genes, and performance also improved when more examples were available to train the models. We demonstrated our predictions using literature evidence of ADAM15, LMNA/C, and DMXL2 genes. All predictions have been implemented in a web resource called "IsoFunc", which is freely available for the global scientific community through http://guanlab.ccmb.med.umich.edu/isofunc .

Metrics for the Human Proteome Project 2016: Progress on Identifying and Characterizing the Human Proteome, Including Post-Translational Modifications

“…It is feasible to predict functional differences between and among splice variants from the same gene using I-TASSER and other protein folding and ligand-binding algorithms. 31 It is also feasible to predict isoform-level functional networks for the different splice isoforms of a given gene such as Her2/neu (ERBB2) using Hisonet 32,33 and IsoFunc 34 from http://guanlab.ccmb.med.umich.edu/.…”

Annotation of Alternatively Spliced Proteins and Transcripts with Protein-Folding Algorithms and Isoform-Level Functional Networks

“…Panwar et al predicted functions for splice isoforms in humans [57]. The approach mainly consists of the following steps:

Download the human RNA-Seq data from the ENCODE study [58]; 127 samples were used.

…”

“…For example, the predictions of human isoform functions [57] and networks [60] are based on RefSeq (version 37.2) and Ensembl (version 74) gene annotations, respectively, so preliminary interpretation of comparative results should be viewed with caution. RefSeq annotation is of high quality but is much less complete compared to Ensembl, which contains many more (predicted) genes and isoforms.…”

Progress in the Chromosome-Centric Human Proteome Project as Highlighted in the Annual Special Issue IV

“…They demonstrate that the simultaneous integration of collision-induced dissociation (CID) and high-energy collisional dissociation (HCD) fragmentation facilitates rapid identification of both glycans and N-glycopeptide backbones in tandem MS data, resulting in the identification of missing proteins having glycans (e.g., Q8N9B8) present in liver cancer (http://gFinder.proteomix.org/). Panwar et al 30 from the Chr 17 team presented a proteome-wide analysis of predicted splice isoform-level functional networks. The accumulation of a large amount of RNA-seq data in the public domain greatly increases our ability to examine the functional annotation of genes at isoform level.…”