Skip to Main Content


Evolutionary simulation of transcription factor-binding site interaction principles

Modeling transcription factor-binding site interactions is of vital importance to enhancing the quality of regulatory network inference algorithms and to improving our understanding transcriptional regulation. In spite of this, the most frequently used model for transcription factor binding (the position-specific weight/scoring matrix, or PSSM) has remained virtually unchanged for over 30 years. Furthermore, given the difficulty of generating large and accurate datasets for multiple transcription factors, it is not obvious what assumptions of this basic model should be relaxed or how. Here we propose to use explicit simulations of the co-evolution of a transcription factor with its target sites in a genomic context to test the validity of different assumptions, such as positional independence, by comparing the evolutionary outcomes obtained with relaxed and constrained models.

Comparative genomics of transcriptional regulatory networks

Comparative genomics is a powerful tool to make inferences on the wiring and evolution of transcriptional regulatory networks, but its application to bacterial regulatory networks is still not well standardized and has been only sparingly used in the analysis of bacterial transcription networks. By leveraging a rapidly-growing amount of experimental data on transcription factor-binding sites, here we seek to standardize comparative genomics analyses of regulatory networks in bacteria and to test their effectiveness for the study of network evolution.


In bacteria, data on transcription factor-binding sites is mostly scattered in model organism-centered databases using different standards and methods. We have developed CollecTF as an open database for transcription-factor binding sites across bacteria. CollecTF ( compiles data on experimentally validated, naturally occurring TF-binding sites across the Bacteria domain, placing a strong emphasis on the transparency of the curation process, the quality and availability of the stored data and fully customizable access to its records.  Furthermore, CollecTF entries are periodically submitted to NCBI for integration into RefSeq complete genome records as db_xref link-out features embedded in genome annotations, to the EBI as regulon information for UniProtKB entries, and as GO annotations through the Gene Ontology Annotation program of the EBI.

Metagenomic analysis of regulatory networks

Next-generation sequencing technologies have made it possible to analyze comprehensively the metagenome of microbial communities. Metagenomes provide an extraordinary amount of sequence data on the genetic composition of a bacterial population. Conventional approaches to the analysis of metagenomes have relied on mapping predicted genes onto known pathways. We have shown that known regulatory data and in silico search methods can be leveraged to reconstruct meta-regulons, and we are currently optimizing the bioinformatics pipeline to allow comparing regulatory networks across metagenomes.

Self-adjustable codon bias indices

Codon usage bias (CUB) is a widespread phenomenon in natural organisms, which depart from a uniform usage of codons (triplets of mRNA letters that designate an amino acid in the genetic code). The genes of many organisms show highly biased codon usages that correlate well with their expression levels. Thus, indices that measure CUB are important tools for prediction of gene expression, optimization of gene sequences or assessment of lateral gene transfer (LGT).