Job offer

We are recruiting a postdoctoral researcher to join the RepliLand project, an interdisciplinary effort at the interface of genomics, artificial intelligence, and biophysics. The project aims to elucidate the mechanisms governing DNA replication initiation in metazoans by combining deep learning, epigenomic data, and mechanistic modelling. The mission is to contribute to the development of predictive models of the replication initiation probability landscape (IPLS) from limited experimental data, with applications in cancer research and cross-species genomics.

Scientific Context.
DNA replication is a fundamental biological process ensuring the faithful transmission of genetic information during cell division. In eukaryotic cells, replication initiates at multiple genomic loci called replication origins, from which two replication forks propagate in opposite directions. The spatial and temporal organization of origin activation defines the replication program, that is both cell-type specific and sensitive to genomic and epigenomic context. The tight regulation of this process remains to be fully understood.

In this context, we developed an original model where the IPLS, the intrinsic propensity of genomic loci to initiate replication, together with fork progression dynamics, fully determines observables of the replication program such as the Mean Replication Timing (MRT) and the Replication Fork Directionality (RFD) profiles. We proposed a strategy to train a neural network to infer high-resolution IPLSs from experimental MRT and RFD data, which when use as an input to the model reproduce experimental replication profiles with unprecedented precision, demonstrating that the IPLS is a powerful and compact representation of the replication program, capturing the key regulatory role of epigenetic and genomic context (Arbona, PLoS Comput. Biol. 2023). However, the application of this approach is currently limited by the scarcity of high-resolution replication datasets, particularly RFD profiles. The project aims to overcome this limitation by developing a new generation of predictive models capable of inferring the IPLS directly from widely available genomic and epigenomic data.

Broadly, the project aims to develop integrative approaches at the interface of (epi)genomics, machine learning and physics, addressing key challenges in modern quantitative biology. The successful candidate will be responsible for:

• Develop and train deep learning models (CNNs, …) data to predict IPLSs from multi-omics datasets (epigenomics, transcription, sequence).
• Using representation learning (tensor factorization framework e.g., AVOCADO) extend the predictive framework allowing imputation in settings with limited data and transfer learning across species from human to mouse, and beyond.
• Apply explainability methods to extract biological insight from trained models.
• Keep up to date with publications on the subject and write scientific articles.

The IBDM, located on the Luminy Campus, comprises around 220 permanent staff - including researchers, lecturers and researchers, engineers and technicians - as well as non-permanent staff on fixed-term contracts, postdocs, PhD students and interns, spread across 22 research teams and 11 technical facilities and services. The IBDM is a joint research unit under the supervision of the CNRS and the AMU, which explores the field of developmental biology and associated pathologies. The work will be carried out within the ‘Computational Biology’ team led by Bianca HABERMANN.

As the laboratory comprises members of various nationalities, fluency in English is essential., * PhD in computational biology, machine learning, bioinformatics or a related field.

• Proficiency with Python programming.
• Proven experience in deep learning (PyTorch or TensorFlow).
• Previous experience with high-dimensional biological data and knowledge in (epi)genomics will be appreciated.
• The ability to work both independently and as part of a team, in an interdisciplinary context at the interface between information sciences and life sciences.
• Strong skills in scientific communication (written and oral); in particular, excellent writing and communication abilities in English are expected