01_RNAseq.Rmd analysis of RNA-seq data
02_ATACseq.Rmd analysis of ATAC-seq data
03_Combined.Rmd joint analysis of TF-gene regulatory networks
RNASEQ_QUANTIFICATION/ data and results used for analysis in 01_RNAseq.Rmd
ATACSEQ/nucleosome_free_regions/ data and results used for analysis in 02_ATACseq.Rmd
plots plots produced in joint analysis of TF-gene regulatory networks
results result of analysis in 03_Combined.Rmd
published published data used for Pou4 target comparisons in 03_Combined.Rmd
Genes were clustered in 9 groups based on their expression fold change in different transgenic lines (Fig2A). Similarly, accessible regions (i.e. ATAC peaks) were also clustered, resulting in 9 groups with accessibility patterns across transgenic lines equivalent to those of gene expression (SFig2). Finally, gene accessibility scores were calculated as weighted sum of the accessibility of peaks assigned to nearby genes, and genes were also clustered in the same 7 clusters based on gene accessibility scores across transgenic lines (Fig2B).
Set of PWMs for archetype motifs was scored across all peaks, and motif hits were determined as those positions in peaks where PWM score was higher than 95th quantile of genomic distribution of motif scores (i.e. motif PWM scores in random genomic positions). Motif enrichment was then calculated for groups of peaks defined earlier (Fig2C).
Footprint scores were calculated for all accessible regions i.e. ATAC peaks (using TOBIAS ATACorrect and TOBIAS FootprintScores). Motif scores and footprint scores were combined (using TOBIAS BINDetect) to determine motif binding in ATAC peaks (Fig2D).
We filtered all genes by expression and targets by footprint score (bound sites from TOBIAS BINDetect) (Fig2D) to construct reporter line-specific Gene Regulatory Networks (GRNs, SFig3). For Pou4, we compared target genes predicted by our computational GRN inference with published ChIP-seq and mutant data (Fig4).