EBI Epigenomic Analysis Pipelines

After an explosion in genomic data, with ever more plant and animal genomes being assembled, research communities are now turning to the dynamics of gene expression regulation. Epigenomic assays, which measure cellular features such as transcription factor binding or chomatin marks, are becoming routine and their field of application has extended beyond human-centric consortia to a diversity of research communities (e.g. modENCODE).

The EMBL-EBI has been a core participant in many of the initial large scale epigenomic consortia (ENCODE, Blueprint, HipSci), and developed for this purpose streamlined processes to store, track, distribute, process and analyse the numerous data files produced by these large scale experiments.

The result of this work is made accessible to users through the Ensembl interface. With a few clicks, a user can request the data from specific experiments, either in raw or processed form. On top of this, we run machine-learning software to detect the regions of the genome that display enhancer, insulator or promoter activity in the available samples.

These pipelines are still being actively improved. The machine-learning step is being updated to integrate the latest algorithms in the field (ChromHMM, Segway). Also, we are developing tools for users to request summary datasets on the fly: instead of accumulating datasets tracks on their screen, a user could collapse them, thus making straightforward comparisons between selected groups.