PhD defence by Xuening He

Genome-wide Rules of Transcription Factor Cooperativity Revealed through Interpretable Deep Learning

Assessment Committee:
Professor Ole Winther, Department of Biology, University of Copenhagen (Chairperson)
Professor Bart Deplancke, EPFL
Associate Professor Jesper Grud Skat Madsen, SDU

Supervisor(s):
Associate Professor Robin Andersson

Department:
Department of Biology

Place:
The defence is conducted as a hybrid defence.

To attend the defence in person:
Biocenter, Building 1, Room: Seminar Room 1.2.03, Ole Maaløes Vej 5, 2200 København

To attend the defence online:
Please follow the link to attend the defence online:
https://ucph-ku.zoom.us/j/62276218689?pwd=Au2rr9XIJR9tpb4wY13TOJeKXmF1Y6.1

MeetingID, if relevant: 622 7621 8689
Password, if relevant: 397564

Email address to gain access to the thesis: xuening.he@bio.ku.dk
You will either receive a copy of the thesis or be informed where you can read a physical copy.

Short description of the thesis:
Transcription factor (TF) cooperativity underlies gene regulation, yet its genomic rules remain poorly defined. We developed DeepCompARE and DeepISA, lightweight convolutional neural network frameworks coupled with in silico ablation (ISA), to quantify motif syntax and TF cooperativity at genome scale. Applied to FANTOM5 CAGE-seq data across 46 cell types, our models reveal that most TF motifs act additively, with deviations captured along a continuum from redundancy to synergy. Redundancy is linked to promoter activity and broad expression, whereas synergy associates with enhancer function, physical TF interactions, and cell-type specificity. Cooperativity scores are robust across motif-scanning stringency, assays, and training stages, establishing generalizable rules of TF regulatory logic.