In silico experiment system for testing hypothesis on gene functions using three condition specific biological networks

•A web-based information system can perform in silico experiments of testing the function of a gene by analyzing miRNA and mRNA data sets measured from knockout mice.•The user hypothesis in a set of English words are converted to genes using our literature and knowledge mining system called BEST.•Co...

Full description

Saved in:
Bibliographic Details
Published in:Methods (San Diego, Calif.) Calif.), 2018-08, Vol.145, p.10-15
Main Authors: Lee, Chai-Jin, Kang, Dongwon, Lee, Sangseon, Lee, Sunwon, Kang, Jaewoo, Kim, Sun
Format: Article
Language:English
Subjects:
Animals
Computational Biology
Computer Simulation
Gene Regulatory Networks
Mice
MicroRNAs - metabolism
Network analysis
Protein Interaction Maps
Transcription Factors - metabolism
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:•A web-based information system can perform in silico experiments of testing the function of a gene by analyzing miRNA and mRNA data sets measured from knockout mice.•The user hypothesis in a set of English words are converted to genes using our literature and knowledge mining system called BEST.•Condition-specific TF, miRNA and PPI networks are automatically generated by projecting gene and miRNA expression data to template networks.•The test result visualizes path from the knockout gene to the target genes through three biological networks. Determining functions of a gene requires time consuming, expensive biological experiments. Scientists can speed up this experimental process if the literature information and biological networks can be adequately provided. In this paper, we present a web-based information system that can perform in silico experiments of computationally testing hypothesis on the function of a gene. A hypothesis that is specified in English by the user is converted to genes using a literature and knowledge mining system called BEST. Condition-specific TF, miRNA and PPI (protein-protein interaction) networks are automatically generated by projecting gene and miRNA expression data to template networks. Then, an in silico experiment is to test how well the target genes are connected from the knockout gene through the condition-specific networks. The test result visualizes path from the knockout gene to the target genes in the three networks. Statistical and information-theoretic scores are provided on the resulting web page to help scientists either accept or reject the hypothesis being tested. Our web-based system was extensively tested using three data sets, such as E2f1, Lrrk2, and Dicer1 knockout data sets. We were able to re-produce gene functions reported in the original research papers. In addition, we comprehensively tested with all disease names in MalaCards as hypothesis to show the effectiveness of our system. Our in silico experiment system can be very useful in suggesting biological mechanisms which can be further tested in vivo or in vitro. Availability: http://biohealth.snu.ac.kr/software/insilico/.
ISSN:1046-2023
1095-9130
DOI:10.1016/j.ymeth.2018.05.003