ChIP-Seq Histone Modification Data

1. H3K4me1 and H3K4me3 data sets

We used ChIP-Seq to characterize genome-wide spatial relationships between histone H3 lysine 4 mono and tri-methylation (H3K4me1, H3K4me3) and transcription factor binding. We generated enrichment profiles for the two histone modifications in unstimulated and interferon-γ stimulated HeLa S3 cells, and used our current methods, and a larger number of reads, to reanalyze the STAT1 ChIP-Seq data we generated for (Robertson 2007 Nature Methods). We compared spatial relationships in this system to those in non-immortalized cells in a different mammal by generating profiles for H3K4me1 and me3 in adult mouse liver and reanalyzing our ChIP-Seq data for Foxa2 binding in this normal tissue (Wederell et al. Nucleic Acids Res. 2008). We describe the data and analysis more fully in Robertson, Bilenky et al., submitted.

2. Reanalyzed datasets for STAT1 and Foxa2 transcription factors

Because we adjusted the number of STAT1 and Foxa2 reads used in the analysis, and applied updated analysis methods, these datafiles differ somewhat from those available at http://www.bcgsc.ca/data/chipseq/ and with the associated publications. We added an extra lane of data for STAT1 binding in both interferon-γ stimulated and unstimulated HeLa S3 cells. For stimulated cells, this increased the number of reads from 15.3 to 18.8 M, and for unstimulated cells from 13.1 to 16.2 M. We reanalyzed the FoxA2 profile using five out of six available lanes of data, working with 11.4 rather than 14.0 M reads.

Analysis methods were as follows. Pooled Eland aligned-read files were filtered to remove gel size ladders and Illumina adapters, duplicate reads were collapsed into single unique reads, an XSET island profile was generated using 200bp virtual fragments, islands were subdivided into enriched regions, and a profile of significantly enriched regions was generated by thresholding islands at a height corresponding to an empirical FDR of ~0.01, using an updated FDR estimation algorithm. The data and analysis are described more fully in Robertson, Bilenky et al., submitted.

Histone lysine trimethylation

Chromatin possesses a multitude of distinct modifications on histones. The precise function of these modifications in transcription is largely unknown but their importance is based on the fact that they correlate with either active or inactive transcription. We used a combination of chromatin immuno-precipitation and a genome-wide next generation DNA sequencing approach (ChIP-seq) to generate a comprehensive set of DNA sequences associated with tri-methylated histones in human promyelocytic leukemia cells (HL60) (Roberston et al., 2007; Barski et al., 2007; Tarjei et al. 2007). We generated more than 294 million reads from 81 Illumina 1G flow cell lanes, yielding more than 7.9 gigabases of equence data. Success in uniquely mapping reads to the reference human genome from each of the 18 ChIP-seq libraries averaged 61%, and resulted in a dataset of more than 181 million uniquely mapped reads.