Benchmarking predictive models of 3-dimensional genome organization

Aim of the Project

Compare the performance of predictive models of chromatin organization and understand which chromatin features define 3D-genome architecture in normal and mutated genomes.

Project overview

We collected the dataset containing capture Hi-C and epigenetics data for wild-type and mutated mouse and human cell types.

Current dataset contains samples from several dozen mouse lines harboring genomic mutations with the known effect on chromatin organization, including data from the groups of Stefan Mundlas, Denis Duboule, Douglas Higgs, Laura Lettice, John Rinn and Narimann Battulin groups. If you have generated 3C-data describing changes of chromatin architecture caused by genetic mutations in human and/or mouse cells, which is not currently in the dataset, please let us know or submit it directly to the dataset.

The data from all sources is now being re-processed to standardize formats and allow uniform comparisons. The uniformly processed data will be next used by participants to generate models/predict effects of genetic mutations on chromatin architecture. Results submitted by participants will be scored according to the predefined accuracy metrics.

Now we are at the last stage of data preprocessing, and we are open to discuss the final pipeline suitable for each invited research group as well as benchmarking criteria. For discussions please join INC-COST slack (channel #benchmarking).

How to contribute

Contact Us

Veniamin Fishman [minja@bionet.nsc.ru]

Genomic mechanisms of Development, Institute of Cytology and Genetics SB RAS