Example applications and input files for OntoFUNC

1. Identify classes of chemicals modulating a pathway

Description: The Chemical Entities of Biological Interest (ChEBI) ontology provides a classification of chemicals based on their structure. Based on known interactions of chemicals with genes or proteins, we are interested in finding classes of chemicals that can modulate a biological pathway, in particular pathways known to be affected in diseases (disease pathways) or pharmacogenomic pathways.

Original dataset: Sympathetic Nerve Pathway (Neuroeffector Junction) and its participants, taken from the Pharmacogenomics Knowledge Base.

Protocol: Identify for a set of chemicals their target genes/proteins. If the chemical targets a gene or protein which is participating in the pathway, the chemical is of interest, otherwise it is not of interest. A detailed description of the method is available here.

OntoFUNC input file: PA2042

OntoFUNC test: hypergeometric test

Ontology: Chemical Entities of Biological Interest (ChEBI)

Results: http://phenomebrowser.net/ontofunc/viewresults.php?id=ontofunc-job-51f1321c981807.67232108

2. Identify abnormal phenotype differences between populations

Description: The Human Phenotype Ontology (HPO) is an ontology characterizing abnormal and clinical phenotypes and can be associated with diseases, e.g., the diseases in OMIM. The Human Disease Network (Hidalgo CA, Blumm N, Barabasi A-L, Christakis NA. PLoS Computational Biology, 5(4):e1000353) contains information on co-morbidity between diseases in different populations. We can use the HPO to identify clinical phenotypes that are significantly different between two populations.

Original dataset: HuDiNe ICD9 5 digit data for Black and White populations.

Protocol: Normalize occurrence data by the number of patients observed in the whole dataset. Identify Human Disease Ontology classes corresponding to ICD codes in HuDiNe. Identify HPO phenotypes for each disease contained in the HuDiNe dataset. Use OntoFUNC over HPO to perform a binomial test comparing phenotypes in both populations.

OntoFUNC input file: black-white.txt

OntoFUNC test: bionomial test

Ontology: Human Phenotype Ontology (HP)

Results: http://phenomebrowser.net/ontofunc/viewresults.php?id=ontofunc-job-51f132946d0cf5.78162754

3. Analysis of gene expression data with the Neuro Behavior Ontology

Description: The Neuro Behavior Ontology (NBO) is an ontology of behavioral processes and phenotypes, extending the behavioral process branch of the Gene Ontology. Annotations to NBO are available from the Rat Genome Database, and several annotations exist for mouse genes. It therefore becomes possible to analyze gene expression datasets using the NBO to reveal detailed information about behavioral differences resulting from differential expression of genes involved in behavioral processes.

Original dataset: Morphine effect on the striatum, Gene Expression Omnibus accession GDS2815

Protocol: We import the dataset into R and perform a t-test to compute differential expression for each probe id between control and chronic morphin use. We subsequently map probe ids to MGI gene identifiers. No correction for multiple testing is performed because the results are intended for use in a Wilcoxon rank test which is based on the ranks of the p-values for differential expression and not on their absolute values or sets of genes that are differentially expressed.

OntoFUNC input file: GDS2815-func-wilcox.txt

OntoFUNC test: Wilcoxon test

Ontology: Neuro Behavior Ontology (NBO)

Results: http://phenomebrowser.net/ontofunc/viewresults.php?id=ontofunc-job-51f131023cf7f0.04033249