Computational protocol: Key regulators control distinct transcriptional programmes in blood progenitor and mast cells

Similar protocols

Protocol publication

[…] ChIP assays were performed as previously described (Forsberg et al, ) using polyclonal antibodies against CTCF (Upstate, 07-729), E2A (Santa Cruz, sc-763x), ERG (Santa Cruz, sc354x), FLI1 (Abcam, ab15289-500), LMO2 (R&D, AF2726), MEIS1 (Santa Cruz, sc-10599x), PU.1 (Santa Cruz, sc-352x), RUNX1 (Abcam, ab23980-100), SCL/TAL1 (Santa Cruz, sc12984x), MITF (Cosmo Bio Co., BAM-73-107-EX), c-FOS (Santa Cruz, sc-253x), H3AcK27 (Abcam, ab4729) and control non-specific rabbit (Sigma, I5006) and goat (Sigma, I5256) IgG. ChIP samples were amplified and sequenced as described (Wilson et al, ). Reads were mapped to the mm9 mouse reference genome using Bowtie (Langmead et al, ) and peaks called using MACS (Zhang et al, ). Mapped reads were converted to density plots and displayed as UCSC genome browser custom tracks.ChIP-Seq peak coordinates from all 10 shared TFs were combined into a single list and peaks overlapping by at least 1 bp were merged. A matrix was generated for peak coverage scores, denoted as cov(HPC-7) and cov(mast), for all the merged coordinates. Coverage scores were counted using the intersectBed function from BEDTools (Quinlan & Hall, ) and then normalised per 10 million reads. For n coordinates and 20 samples, this produces an n × 20 coverage matrix. Hierarchical clustering and heatmap of the Pearson correlation coefficient between each pair of datasets were generated in R. This matrix was used for further analysis—differential score calculation and regression modelling. [...] De novo motif analysis on the central 100-bp peak regions was carried out using the Homer software (Heinz et al, ), and matches to known motifs were discovered using TOMTOM (Gupta et al, ). Similarity to known motifs was discovered from 2 databases—Jaspar and UniProbe (Bryne et al, ; Robasky & Bulyk, ). Only significant motifs (q-value ≤ 0.01) were reported. To carry out the motif content analysis, the same procedure as above was carried out on cell type-specific and common peak regions. Peak regions for each of the 10 shared TFs were divided into three categories: HPC7-specific (HPC7-bound only), mast cell-specific (mast-bound only) and common (HPC7/mast peaks overlapping ≥ 1 bp) regions. Results from de novo motif analysis were combined in a matrix where columns are datasets, rows are Jaspar/UniProbe motifs and each row/column value is x or 0, where x corresponds to the fraction of peaks with significant (TOMTOM q-value ≤ 0.05) match to a particular motif, while 0 is a non-significant match. The matrix was used to generate a heatmap that indicates fraction of peaks containing motif in cell type-specific and common regions. [...] Bar charts and scatterplots were plotted using ggplot2 (Wickham, ). Venn diagrams were created using the venneuler package (Wilkinson, ). Heatmaps were generated using the gplots packages (Warnes et al, ). […]

Pipeline specifications

Software tools HOMER, Tomtom, Ggplot2, venneuler, gplots
Databases UniPROBE
Applications Miscellaneous, ChIP-seq analysis
Organisms Homo sapiens