We present a physics-based machine learning approach to predict transcription factor binding affinities in vitro from structural and mechanical DNA properties directly derived from atomistic molecular dynamics simulations.

The method is able to predict affinities obtained with techniques as different as uPBM, gcPBM and HT-SELEX with a predictive power superior to existing algorithms and a full generality which can be used to in the future to accommodate non-coding nucleobases.

When complemented with chromatin structure information, our in vitro trained method provides estimates of in vivo binding sites in yeast with unprecedented accuracy.

[maxbutton id=”4″ url=”https://mmb.irbbarcelona.org/bioexcel-docs/data/BioExcel2_UC3_WF4/pdf” text=”Preprint” ]


Sandro Barissi, Alba Sala, Milosz Wieczor, Federica Battistini, Modesto Orozco (2022):
DNAffinity: A machine-learning approach to predict DNA binding affinities of transcription factors.

About the author

Stian works in School of Computer Science, at the University of Manchester in Carole Goble‘s eScience Lab as a technical software architect and researcher. In addition to BioExcel, Stian’s involvements include Open PHACTS (pharmacological data warehouse), Common Workflow Language (CWL), Apache Taverna (scientific workflow system), Linked Data and identifiers, research objects (open science) and digital preservation, myExperiment (sharing scientific workflows), provenance (where did things come from and who did it) and annotations (who said what). orcid.org/0000-0001-9842-9718