This repository contains the use example for the Chicdiff package for the differential analysis of Promoter Capture Hi-C data. Refer to the Chicdiff manuscript (Cairns/Orchard/Malysheva & Spivakov, bioRxiv preprint https://doi.org/10.1101/526269) and the package home on Github (https://github.com/RegulatoryGenomicsGroup/chicdiff) for more details.
The use example is based on the Promoter Capture Hi-C datasets for three replicates of monocytes and four replicates of CD4+ T human cells from Javierre et al., Cell 2016.
Javierre_ind_rep_peak_matrix.txt contains the list of interactions of interest (having a Chicago score of 5 in at least one replicate) that were submitted to Chicdiff testing.
*RDa files contain the output of the Chicago package for each replicate.
*len_distSign.txt files contain the interaction-level read count data for each replicate in the "chinput" format.
DesignDir.tar.gz is an archive containing the design files that needs to be decompressed into a "design directory" needed to run Chicdiff on these data.
GeneExpressionMatrix.txt contains MMSEQ-processed expression data from Javierre et al., including those for monocytes and CD4+ T cells, which were used in the analysis.
Results in the paper are based on four Chicdiff runs on these data with different parameters:
- chicdiff_run_CD4_vs_Mono_unmerged.R (default normalisation [norm="combined"] and region extension [RUexpand=5] settings).
- chicdiff_run_CD4_vs_Mono_unmerged_RU0.R (default normalisation, but no region extension [RUexpand=0]).
- chicdiff_run_CD4_vs_Mono_unmerged_DEseq.R (DESeq2 normalisation [norm="standard"] and default region extension [RUexpand=5]).
- chicdiff_run_CD4_vs_Mono_unmerged_DEseq_RU0.R (DESeq2 normalisation and no region extension).
Each of these scripts was run separately on a Linux cluster with ~100 Gb RAM allocated for each run.
Data for table S1 and figures were generated with the code in use_example_analysis_final.R based on the Chicdiff results generated as above and the expression matrix.
