Functional Annotation of Human Protein Coding Isoforms via Non-convex Multi-Instance Learning

Luo, Tingjin; Zhang, Weizhong; Qiu, Shang; Yang, Yang; Yi, Dongyun; Wang, Guangtao; Ye, Jieping; Wang, Jie

doi:10.1145/3097983.3097984

Cited by 19 publications

(41 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Therefore, these cannot be directly applied to mRNA isoform function prediction because they ignore the distinct functions of alternatively spliced mRNA isoforms. However, important advancements have been made by recent studies towards mRNA isoform level understanding of gene functions [18,19,[21][22][23][24][25] such as the prediction of more immune related gene ontology terms for the mRNA isoform ADAM15B than isoform ADAM15A of ADAM15 gene, which is involved in B-cell-mediated immune mechanisms.…”

Section: Introductionmentioning

confidence: 99%

Tissue-specific mouse mRNA isoform networks

Kandoi

Dickerson

2019

Preprint

View full text Add to dashboard Cite

Alternative Splicing produces multiple mRNA isoforms of genes which have important diverse roles such as regulation of gene expression, human heritable diseases, and response to environmental stresses. However, little has been done to assign functions at the mRNA isoform level. Functional networks, where the interactions are quantified by their probability of being involved in the same biological process are typically generated at the gene level. We use a diverse array of tissue-specific RNA-seq datasets and sequence information to train random forest models that predict the functional networks. Since there is no mRNA isoform-level gold standard, we use single isoform genes co-annotated to Gene Ontology biological process annotations, Kyoto Encyclopedia of Genes and Genomes pathways, BioCyc pathways and protein-protein interactions as functionally related (positive pair). To generate the non-functional pairs (negative pair), we use the Gene Ontology annotations tagged with "NOT" qualifier. We describe 17 Tissue-spEcific mrNa iSoform functIOnal Networks (TENSION) following a leave-one-tissue-out strategy in addition to an organism level reference functional network for mouse. We validate our predictions by comparing its performance with previous methods, randomized positive and negative class labels, updated Gene Ontology annotations, and by literature evidence. We demonstrate the ability of our networks to reveal tissue-specific functional differences of the isoforms of the same genes.

show abstract

Section: Introductionmentioning

confidence: 99%

Tissue-specific mouse mRNA isoform networks

Kandoi

Dickerson

2019

Preprint

View full text Add to dashboard Cite

show abstract

“…There are many computational approaches that provide the mechanism that allows for the prediction of the instance or instances of interest that are defined by an MILtype problem. Some approaches employ mathematical, or statistical methods such as logistical regression, or maximum likelihood, or more machine learning based methods such as SVM [31][36][26] [2]. This machine learning technique has the ability to solve problems in many fields where there is a hierarchy between the pieces of the problem to be predicted or when there is a parent child relationship.…”

Section: Mil-based Methodsmentioning

confidence: 99%

“…In the case of alternative splicing isoform products functional annotations, computational methods have been integral in advancing our understanding of them [31] [36][12] [54].…”

Section: Computational Approachesmentioning

confidence: 99%

“…In addition, some research questions such as the degree of functional diversification that AS exerts based on the organism type, or the different roles AS may play in the process of evolution are examples of questions that need further investigation [56]. Also, when it comes to investigating functions at the isoform level, no golden function-labeling standard for isoforms exists which makes it difficult for researchers to annotate functions to isoforms [31] [36].…”

Section: Introduction 11 Background and Motivationmentioning

confidence: 99%

“…Similar to micro-array data, RNA-seq covers different types of tissues and cell linings; however, RNA-seq is capable of detecting expression data at the isoform level more efficiently than micro-arrays [22]. Deep sequencing of RNA transcripts in tissue samples provides a wealth of data to study AS and to be able to differentiate isoforms based on expression levels [31]. Many methods have been developed to precisely measure the expression levels for each transcript; thus, allowing for the transfer of functional annotations from the gene level to the isoform level.…”

Section: Introduction 11 Background and Motivationmentioning

confidence: 99%

See 2 more Smart Citations

Analysis of Differences Between Human Alternative Splicing Protein Isoforms and Their Links to Diseases

Fallatah¹

View full text Add to dashboard Cite

Alternative Splicing (AS) is a process that is believed to have links to cellular function changes and some diseases in humans. Although AS was first discovered in the 1970s, not much research has been conducted on its role in functional implications on the proteome level. This study aims to use PIPE, a protein-protein interaction prediction algorithm, along with a tissue expression dataset to build a pipeline that differentiates between AS isoform products by analyzing isoform sequence changes, functional changes, and tissue expression changes that AS introduces. The study found that isoform sequence changes in alternative isoforms tend to be conserved deletions of amino-acid sub-sequences. The study also found that there is a statistically significant overlap between PIPE-predicted protein-protein interaction (PPI) network changes and tissue expression changes of alternatively spliced isoforms (ASIs) relative to their canonical isoforms (CIs) with a p-value of 8.25×10 −5 . Finally, among the analysis pipeline top ten genes with predicted significant ASIs' PPI network changes, LMO2, THOC2, and UBE2L3 are genes that were suspected of having links to different diseases such as basel-type breast cancer, intellectual disability (ID) and numerous autoimmune diseases according to literature studies. i Firstly, I would like to thank my supervisor Dr. Frank Dehne for his outstanding support throughout my Master's degree. Dr. Dehne provided me with invaluable advice, support, and kindness during my Master's journey. I could not have asked for a better supervisor.

show abstract

Imbalance deep multi‐instance learning for predicting isoform–isoform interactions

Zeng

Wang

et al. 2021

Int J Intell Syst

View full text Add to dashboard Cite

Multi‐instance learning (MIL) can model complex bags (samples) that are further made of diverse instances (subsamples). In typical MIL, the labels of bags are known while those of individual instances are unknown and to be specified. In this paper we propose an imbalanced deep multi‐instance learning approach (IDMIL‐III) and apply it to predict genome‐wide isoform–isoform interactions (IIIs). This prediction task is crucial for precisely understanding the interactome between proteoforms and to reveal their functional diversity. The current solutions typically formulate the prediction of IIIs as a MIL problem by pairing two genes as a “bag” and any two isoforms spliced from these two genes as “instances.” The key instances (interacting isoform pairs) trigger the label of the positive (interacting) gene bags, which is important for identifying the IIIs. Furthermore, the prediction task was simplified as a balanced classification problem, which in practice is a rather imbalanced one. To address these issues, IDMIL‐III fuses RNA‐seq, nucleotide sequence, amino acid sequence and exon array data, and further introduces a novel loss function to separately model the loss of positive pairs and of negative pairs, and thus to avoid the expected loss dominated by majority negative pairs. In addition, it includes an attention strategy to identify positive isoform pairs from a positive gene bag. Extensive experimental results prove the effectiveness of IDMIL‐III on predicting IIIs. Particularly, IDMIL‐III achieves an F1 value as 95.4%, at least 3.8% higher than those of competitive methods at the gene‐level; and obtains an F1 as 29.8%, at least 2.4% higher than the state‐of‐the‐art methods at the isoform‐level. The code of IDMIL‐III is available at http://mlda.swu.edu.cn/codes.php?name=IDMIL-III.

show abstract

Functional Annotation of Human Protein Coding Isoforms via Non-convex Multi-Instance Learning

Cited by 19 publications

References 31 publications

Tissue-specific mouse mRNA isoform networks

Tissue-specific mouse mRNA isoform networks

Analysis of Differences Between Human Alternative Splicing Protein Isoforms and Their Links to Diseases

Imbalance deep multi‐instance learning for predicting isoform–isoform interactions

Contact Info

Product

Resources

About