Differential principal component analysis of ChIP-seq

Abstract

We propose differential principal component analysis (dPCA) for analyzing multiple ChIP- sequencing datasets to identify differential protein–DNA interactions between two biological conditions. dPCA integrates unsupervised pattern discovery, dimension reduction, and statistical inference into a single framework. It uses a small number of principal components to summarize concisely the major multiprotein synergistic differential patterns between the two conditions. For each pattern, it detects and prioritizes differential genomic loci by comparing the between-condition differences with the within-condition variation among replicate samples. dPCA provides a unique tool for efficiently analyzing large amounts of ChIP-sequencing data to study dynamic changes of gene regulation across different biological conditions. We demonstrate this approach through analyses of differential chromatin patterns at transcription factor binding sites and promoters as well as allele-specific protein–DNA interactions.

Developmental exposure to the organophosphorus flame retardant tris(1,3-dichloro-2-propyl) phosphate: Estrogenic activity, endocrine disruption and reproductive effects on zebrafish

“…This result indicates that parental exposure to TDCPP causes developmental toxicity in offspring. Parental exposure to toxicants, including estrogenic chemicals, have frequently been associated with an increased malformation rate in F1 eggs/larvae, including pentachlorophenol (Ma et al, 2012), polybrominated diphenyl ethers (Yu et al, 2011;Chen et al, 2012) bisphenol S (Ji et al, 2013), benzo[a]pyrene (Corrales et al, 2014) and natural mixture of persistent organic pollutants (Lyche et al, 2013). Taken together, these data indicate that long-term exposure to lower concentrations of TDCPP in parental fish causes developing toxicity in the F1 generation.…”

“…However, production of steroids is a complex process with multiple sensitive control points, so the genes within the steroidogenesis pathway are not transcribed to the same extent and simple statistical correlations between mRNA expression and hormone production may not be expected (Gracia et al, 2007). To clarify the relationship between female steroid hormone levels and fecundity, a PCA was used to reduce the dimension of gene transcriptional profiles by transforming them into a small number of PCs (Ji et al, 2013). In females, PC1, which was highly influenced by variables such as fshˇ, lhˇ, gnrh2, lhr, Activin-ˇ2A, er˛, erˇ, vtg1 and vtg3 was positively correlated with the concentrations of E 2 and T, while negatively correlated with fecundity.…”

An Integrated Model of Multiple-Condition ChIP-Seq Data Reveals Predeterminants of Cdx2 Binding

“…Other multi-condition approaches focus on ChIP-seq signals arising from broad regions of enrichment, such as histone modifications. These methods instead search for larger genomic regions where coverage patterns differ across experiments [8,[13][14][15]. In contrast, MultiGPS uses a joint multi-experiment model that considers the read data from all experiments to produce accurate location estimates of punctate binding events.…”

Quantitative Comparison of ChIP-Seq Data Sets Using MAnorm

“…Furthermore, transcription factors and epigenetic modifications usually act together to modulate gene expression (Bernstein et al 2007). Most recently, statistical models have been developed to study such combinatorial patterns in a genome-wide fashion (Pinello et al 2014;Kasowski et al 2013;Bernstein et al 2007;Ji et al 2013). However, although people have widely found changes of epigenetic marks and transcriptional factors' binding often correlate with other (Shao et al 2012;Bernstein et al 2007), quite few computational models were developed to dissect how they collectively define the differential expression program between different cell types (Ji et al 2013).…”

“…Most recently, statistical models have been developed to study such combinatorial patterns in a genome-wide fashion (Pinello et al 2014;Kasowski et al 2013;Bernstein et al 2007;Ji et al 2013). However, although people have widely found changes of epigenetic marks and transcriptional factors' binding often correlate with other (Shao et al 2012;Bernstein et al 2007), quite few computational models were developed to dissect how they collectively define the differential expression program between different cell types (Ji et al 2013). Here we used a systematic comparison of associated histone marks at distal active enhancers between adult and fetal stages of human erythroid cells as example, to illustrate how to design downstream applications of MAnorm for integrative analyses.…”