Description of the thesis topic

With this PhD project, we propose to construct machine learning models capable of deciphering the epigenomic regulation of breast tumors, from the healthy mammary gland to tumors treated with chemotherapies. To do so, an inhouse single cell epigenomic dataset of mouse and human mammary gland (unpublished) will be used. These datasets will also be completed with public DNA accessibility measure. We will first train a deep learning classifier model to recognize cell types composing the healthy mammary gland. Then, using transfer learning methods, we will train the model into identifying tumor cells, both naive and treated with chemotherapies. Finally, we will use the model to do in silico evolution to identify key epigenetic changes responsible for tumorigenesis and chemotherapies tolerance. This project will be led in collaboration with wet lab biologists of the group for data production and also validation of model predictions in breast models (cell lines or organoids).

Work Context

The Vallot Lab is a dynamic & creative research group in UMR3244/Translational Department, Institut Curie (Paris, FR). Our lab aims in understanding epigenomic tumor evolution and leverage its reversibility to enhance response to standard of cares and intercept tumor initiation. We have established a unique expertise in cancer epigenetics, single-cell epigenetics and RNA sequencing, and bioinformatics analyses.
We have recently developed single cell epigenomic wet lab and computational methods and generated several multi-modal epigenomic datasets studying the mammary gland in normal and tumor states. We are lucky to be embedded near hospital, which favours translational discoveries.