Premium
Bundled workflow of Predictive Soft‐wares for Biomarker Identification and cross‐validation
Author(s) -
Muhie Seid,
Hammamieh Rasha,
Jett Marti
Publication year - 2013
Publication title -
the faseb journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.709
H-Index - 277
eISSN - 1530-6860
pISSN - 0892-6638
DOI - 10.1096/fasebj.27.1_supplement.lb139
Subject(s) - random forest , identification (biology) , workflow , cross validation , computer science , machine learning , artificial intelligence , predictive modelling , biomarker , binary classification , data mining , biology , database , support vector machine , biochemistry , botany
Choosing the right predictive tool for identification of molecular biomarkers from noisy high throughput data is a challenge and often leads to unstable predictive features which may be no better than random selection. The use of an ensemble of predictors (such as random forest) has led to an improvement of the likelihood of a hit to be truly predictive. We bundled together different ensembles of prediction and classification tools (including random forest) for both identification and cross‐validation of predictors or classifiers. Our tool is assembled and implemented in R programming language. The user has the option to choose the number and combination of ensembles to identify cross‐validated potential biomarkers along with appropriate scores. Using our tool, we have identified Staphylococcal enterotoxin B‐exposure predictors (transcripts) from publicly available gene expression omnibus data sets (GDS 3399, GSE 4478, GSE 13738, and GSE 15571). Indentified transcripts were cross‐validated and found to be specific to SEB exposure under different pathophysiologic conditions.