The project aims at identifying the interactions between regulatory proteins and genomic DNA sequences at a large-scale and to generate expert systems for the modeling and prediction of these interactions using bioinformatics.
We contribute to the development of genomic methods for high throughput detection of binding sites of regulatory proteins on DNA. These approaches can quickly measure the affinity of transcription factors for a collection of sequences of DNA in vitro, so as to correlate these measures with the regulatory function of these sequences in vivo. These data are also used to predict regulatory genomic sequences using bioinformatics tools generated by our collaborators at the Swiss Institute of Bioinformatics (P. Bucher and collaborators, C. Mazza and collaborators). Our work has shown that some models can accurately predict the binding affinity of transcription factors for any DNA sequence.
Studies of these interactions by DNA microarrays (protein-binding microarrays) have shown that it is possible to test quickly the activity of regulatory proteins responsible of carcinogenesis and of tumor resistance to pharmacological treatment. Our work has shown that this approach is applicable to the molecular study of biopsies of breast cancer, providing an analytical method to study molecular markers of tumor progression. In collaboration with researchers and physicians of CHUV, we evaluate some of these interactions as a success predictor of neoadjuvant chemotherapy, to improve the diagnosis of breast cancer and the choice of better therapeutic approaches.