Transcription Factor-Interaction Prediction and Classification

Participants: Stefano Perna, Pietro Pinoli, Stefano Ceri, Wong Limsoon

Background

Transcriptional regulation is critical to the cellular processes of all organisms. Regulatory mechanisms often involve more than one transcription factor (TF) from different families, binding together and attaching to the DNA as a single complex, but only a fraction of the regulation partners of each transcription factor is currently known. It is thus of interest to biologists to analyse data extracted from TF-DNA binding experiments (e.g. ChIP-Seq) to infer the mechanisms that drive transcriptional regulation. In this project, we developed techniques for predicting the physical interaction between transcription factors, as well as for predicting the nature of their interactions (i.e. co-operative, competitive, or others).

Achievements

• The TICA method and the associated TICA web server for predicting TF-TF interactions. Its main novelty was the use of a form of non-averaging statistics, viz. the size of the tail of the distance distribution between nearest binding sites of the candidate pair at the promoters of candidate target regulated genes in specific cell lines. There was a good complementarity between the factors considered by TICA and two competing methods (TACO and CENTDIST) such that a simple union of the three’s predictions delivered more than 10% improvement in the geometric mean of their individual recall and specificity.

• The NAUTICA method which employed information from protein-protein interaction (PPI) networks to assign TF-TF interaction candidates to one of three classes: competition, co-operation, and non-interaction. NAUTICA worked well in separating co-operating from competing TFs, which was of interest since there was no other computational method that could make this distinction.

• We demonstrated inconsistencies between two main sources of TF binding data, Cistrome and ENCODE. We discovered that consistency would be improved 3-fold by restricting Cistrome to high-intensity peaks (top 25%). Using the intersection of Cistrome high-intensity and ENCODE as input, TF-TF interaction prediction using e.g. TICA would improve significantly.

Selected Publications

• Stefano Perna, Arif Canakoglu, Pietro Pinoli, Stefano Ceri, Limsoon Wong. Implementing a transcription factor interaction prediction system using the GenoMetric Query Language. Data Mining for Systems Biology: Methods and Protocols, edited by Hiroshi Mamitsuka, Chapter 6, pages 63--81, Springer, July 2018.

• Stefano Perna, Pietro Pinoli, Stefano Ceri, Limsoon Wong. TICA: Transcription interaction and coregulation analyser. Genomics, Proteomics, & Bioinformatics, 16(5):342--353, October 2018. PDF

• Stefano Perna, Pietro Pinoli, Stefano Ceri, Limsoon Wong. NAUTICA: Classifying transcription factor interactions by positional and protein-protein interaction information. Biology Direct, 15:13, September 2020. PDF

• Stefano Perna, Pietro Pinoli, Stefano Ceri, Limsoon Wong. A comparative analysis of ENCODE and Cistrome in the context of TF binding signal. BMC Genomics, 25(Suppl 3):817, August 2024. PDF

Dissertations

• Stefano Perna, "Data-driven techniques for knowledge discovery in regulomics", PhD thesis, Politecnico di Milano, Italy, 2019

Acknowledgements