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

Abstract

Regulatory proteins can bind to different sets of genomic targets in various cell types or conditions. To reliably characterize such condition-specific regulatory binding we introduce MultiGPS, an integrated machine learning approach for the analysis of multiple related ChIP-seq experiments. MultiGPS is based on a generalized Expectation Maximization framework that shares information across multiple experiments for binding event discovery. We demonstrate that our framework enables the simultaneous modeling of sparse condition-specific binding changes, sequence dependence, and replicate-specific noise sources. MultiGPS encourages consistency in reported binding event locations across multiple-condition ChIP-seq datasets and provides accurate estimation of ChIP enrichment levels at each event. MultiGPS's multi-experiment modeling approach thus provides a reliable platform for detecting differential binding enrichment across experimental conditions. We demonstrate the advantages of MultiGPS with an analysis of Cdx2 binding in three distinct developmental contexts. By accurately characterizing condition-specific Cdx2 binding, MultiGPS enables novel insight into the mechanistic basis of Cdx2 site selectivity. Specifically, the condition-specific Cdx2 sites characterized by MultiGPS are highly associated with preexisting genomic context, suggesting that such sites are pre-determined by cell-specific regulatory architecture. However, MultiGPS-defined condition-independent sites are not predicted by pre-existing regulatory signals, suggesting that Cdx2 can bind to a subset of locations regardless of genomic environment. A summary of this paper appears in the proceedings Competing Interests: The authors have declared that no competing interests exist. * E-mail: mahony@psu.edu (SM); dkg@mit.edu (DKG) . These authors contributed equally to this work.

The Lineage-Specific Transcription Factor CDX2 Navigates Dynamic Chromatin to Control Distinct Stages of Intestine Development

“…Our experiments with forced CDX2 expression in hESC-derived endoderm are similar to our findings in the mouse, with the observation CDX2 is not sufficient to access at many of its primary hindgut targets independently of the signaling environment. These findings are consistent with CDX2's ability to pioneer certain binding sites in ES cells, while depending on accessible chromatin at others (Mahony et al, 2014) Figure 4B). CDX2 binding during intestinal specification is likely to rely upon pioneer factors, and motif analysis points to FoxA, a known pioneer factor (McPherson et al, 1993), as a likely candidate ( Figure S3C).…”

Differential Principal Components Reveal Patterns of Differentiation in Case/Control Studies

“…As a result, their proposed method first computes the mean of the data samples in each group, and then performs PCA on the dataset of differences between means. This process leads to selection of axes which characterize the protein-binding process studied in CHiP-Seq data, and has been successfully applied in several studies (5,6). When applying this idea to settings beyond ChIP-seq data, we encounter a major problem: the number of principal components is strictly less than the number of datapoints.…”

Alignment and quantification of ChIP-exo crosslinking patterns reveal the spatial organization of protein-DNA complexes

“…If the maximization step results in two components sharing the same positions, they are combined in the next iteration of the algorithm. Crosslinking positions are simultaneously modeled across all analyzed ChIP-exo experiments in a given aligned profile, and information about crosslinking position estimates are shared between experiments at each EM step using a positional prior strategy described in the MultiGPS algorithm (Mahony et al, 2014). This positional prior encourages consistency in estimated crosslinking positions across ChIP-exo experiments.…”