Toxicogenomic profiles of neuronal targeting insecticides in zebrafish embryos as non-target aquatic vertebrate model

We have conducted semi-static exposure studies with six neuronal targeting insecticides on fertilized zebrafish (Danio rerio) eggs, similar to the OECD 236 guideline for the 96h zebrafish embryo toxicity test. The aim of these transcriptomic profiling experiments was to screen for ecotoxicogenomic fingerprints in zebrafish (Danio rerio) embryos as aquatic vertebrate non-target model exposed to sub lethal concentrations of pesticides. Data published in Reinwald et al. 2021 (PMID:34748799).

For experimental details please refer to the publicly accessible experiment description and treatment protocols deposited in the ArrayExpress database at EMBL-EBI (www.ebi.ac.uk/arrayexpress) under the following accession numbers: E-MTAB-9852 (Abamectin), E-MTAB-9855 (Carbaryl), E-MTAB-9853 (Chlorpyrifos), E-MTAB-9854 (Fipronil), E-MTAB-9859 (Imidacloprid), E-MTAB-9860 (Methoxychlor).

The uploaded data archives (7-zip compressed) consists of three major data types:
1. MultiQC reports from raw RNA-Seq read processing and QC (50bp SR) ( Neuotox_multiQCreports.7z )
2. Result tables of differential gene expression analysis (DGEA) with DESEq2 ( Neurotox_DESeq2_ResultTables.7z )
3. Result tables of gene set enrichment analysis (GSEA) with clusterProfiler ( Neurotox_clusterProfiler_ResultTables.7z )
4. Result tables of overrepresentation analysis (ORA) via clusterProfiler::compareCluster() ( Neurotox_clusterProfiler_ORA_on_core_DEGs.7z )

Each data archive contains a README file describing the methods applied to generate the respective result tables / reports. For each tested substance and exposure condition a DGEA and GSEA result table is uploaded. The corresponding bash and R codes for each analysis step are available on github under:
https://github.com/hreinwal/zfeNeurotox

Gene count normalization and DGEA was conducted with DESeq2 (Love et al., 2014, DOI 10.1186/s13059-014-0550-8). Three biological replicates per condition, exposure treatments were compared with respect to the control group in a pairwise fashion, applying Wald’s t-test. P values were corrected for multiple testing with independent hypothesis weighting (IHW) (Ignatiadis et al., 2016, DOI 10.1038/nmeth.3885) after Benjamini-Hochberg (BH). To improve the signal to statistical noise ratio, the obtained log2-fold change (lfc) values were shrunk with the apeglm method described by Zhu and colleagues (2019, DOI 10.1093/bioinformatics/bty895) before DGEA result tables were subjected to GSEA via clusterProfiler (Yu et al., 2012, DOI 10.1089/omi.2011.0118)
and reactomePA (Yu and He, 2016, DOI 10.1039/C5MB00663E). The linked ArrayExpress accession numbers above, provide access to the raw and DESeq2 normalized gene count matrices upon which these analysis were performed. Genes were annotated through the biomaRt package (Durinck et al., 2009, DOI 10.1038/nprot.2009.97) in R (R Core Team 2021).