Researchers at TUT develop the most accurate model for protein binding prediction

The model is based on data obtained from microarray-based protein binding measurements.The model sheds light on the mechanisms that regulate cell functions and can benefit, for example, cancer research.

A paper that compares the currently available methods for predicting the binding affinity of proteins to specific DNA sequences was recently published in Nature Biotechnology. After a total of 26 methods were evaluated, a computational model developed by a team of researchers at the Department of Signal Processing of Tampere University of Technology (TUT) was found to outperform the others. The model also won the international DREAM5 challenge in 2010. 

“Our solution uses a linear model to model binding affinities from multiple short DNA sequences. This approach produces more accurate results than conventional statistical models depicting the protein binding site,” says Matti Annala. In addition to Annala, researchers Kirsti Laurila, Matti Nykter and Harri Lähdesmäki were involved in developing the model.   

The paper published in Nature Biotechnology compares the models submitted to the DREAM5 challenge to similar models that are currently available. The comparisons were conducted in collaboration between the organizers of DREAM5 and the teams that took part in the challenge.    

Implications for multiple areas of biological and clinical research

DREAM stands for Dialogue for Reverse Engineering Assessments and Methods. Each year, DREAM brings together researchers from around the world to tackle different challenges relating to computational systems biology and gene regulatory networks.

“The DREAM challenges aim to clearly identify the strengths and weaknesses of different prediction models and determine their overall reliability,” says Matti Nykter, who currently holds a professorship at the University of Tampere. 

The researchers who took part in the DREAM5 challenge in 2010 were assigned to develop a computational model for predicting the DNA binding preferences of transcription factors (TFs). TFs are proteins that bind to specific DNA sequences and are essential for the regulation of gene expression. If there are accurate protein-DNA binding models available, researchers will be in a better position to understand cellular regulatory mechanisms and thereby contribute to the fight against cancer and other diseases. Breakthroughs in modelling the binding preferences of TFs will have implications for both biological and clinical research. Gaining more insight into the mechanisms underlying the behaviour of DNA-binding proteins is especially relevant for cancer research. Mutations in TFs result in abnormal cell functions and have been associated with several types of cancer.   

The DREAM5 challenge was arranged by Columbia University and IBM and sponsored by the U.S National Institute of Health.

Further information:

Tampere University of Technology
Research Assistant Matti Annala, tel. +358 40 198 1315, ">matti.annala@tut.fi

University of Tampere
Professor Matti Nykter, tel. +358 40 849 0651, ">matti.nykter@uta.fi

The findings were published in Nature Biotechnology.