Computational protocol: Genome-Wide Gene Expression Analysis in Response to Organophosphorus Pesticide Chlorpyrifos and Diazinon in C. elegans

Similar protocols

Protocol publication

[…] A Perking & Elmer scanner was used to extract the raw intensities from the microarrays. Normalization within arrays and normalization between arrays of raw intensities was done using loess method and aquantile method , respectively. Both methods are included in the Limma package from R software (www.r-project.org/). The Rank Product package was used to identify the differentially expressed genes between controls and treatment in each experiment. Briefly, genes were ranked based on up- or downregulation by the treatment in each experiment. Then, for each gene a combined probability was calculated as a rank product (RP). The RP values were used to rank the genes based on how likely it was to observe them by chance at that particular position on the list of differentially expressed genes. The RP can be interpreted as a p-value. To determine significance levels, the RP method uses a permutation-based estimation procedure to transform the p-value into an e-value that addresses the multiple testing problem derived from testing many genes simultaneously. Genes with a percentage of false-positives (PFP) <0.05 were considered differentially expressed between treatments and control in each experiment.This method has the advantage to identify genes with a response to the toxicants even when the absolute effect of the response was low. Because we used sub-lethal concentrations of the toxicants, methods that use thresholds based on absolute fold change would not identify small changes in gene expression. Moreover, RP has proved to be a robust method for comparing microarray data from different sources and experiments .Gene Ontology (GO) data and functional domain data were extracted from Wormbase release WB195 using the R package BioMart . GO terms and domains with less than 4 genes were discarded. Over-represented groups of GO terms and functional domains were identified using a hypergeometric test, with a threshold of p-value<0.01. The hypergeometric test compared a group of 396 GO terms, with 16,947 annotated genes, with the GO terms associated with the significantly regulated genes in each treatment (551,245 and 233 for CPF, DZN and the LDM). For functional domains analysis, 1003 InterPro and Pfam domain terms were used, with 8682 annotated genes.The same hypergeometric test was used to determine significant regulation of the different pathways analyzed. The lists of regulated genes, in this case, were extracted from the original publications. Groups sizes were 541 genes for the daf-16 pathway, 255 for the etl-2 pathway, 305 for the mdt-5 pathway, 320 for the dbl-1 pathway, 143 for the pmk-1 pathway, and 290 genes for Cd responsive genes. Annotated genes were all the unique genes on the microarray, 18893.Microarray data have been deposited in Gene Expression Omnibus (www.ncbi.nlm.nih.gov/geo/), accession number GSE16719. […]

Pipeline specifications

Software tools limma, BioMart
Databases WormBase
Application Genome annotation
Organisms Caenorhabditis elegans
Diseases Drug-Related Side Effects and Adverse Reactions
Chemicals Acetylcholine, Diazinon, Chlorpyrifos