Title : postdoc / data scientist position on graph-representation learning and single-cell data analysis for therapy optimization in cancer

Job description : cell therapy, embodied by the success of chimeric antigen receptor (CAR)T cells in oncology, has the potential to revolutionize the treatment of multiple diseases. Among the many challenges that are slowing down their adoption, turning cell therapies from research tools to drug products is the most pressing one. Manufacturing CAR T cells is a complex procedure that depends on extrinsic (culture conditions) as well as intrinsic (patient-specific) factors. To better understand key features driving the success of this procedure, classical cytometry methods are increasingly supplemented by high-dimensional methods such as CITEseq, where the expression of thousands of genes and tens of proteins can be measured at once at the single cell level.

While those layers of data are usually treated as distinct during analysis, they are heavily correlated as genes are translated into proteins, that in turn influence the expression of genes. Capturing those complex relationships, similarly to the combined analysis of scRNAseq and scATACseq data, would help refine our analysis and support the identification of new cell types.

In the context of a collaboration with Janssen Pharmaceutica, the Clinical Bioinformatics lab of the Imagine Institute is seeking a postdoc / data scientist to participate in the development of innovative multi-omics integration methods, and their application to ongoing drug discovery projects. The research will be performed in close collaboration with the AI for Single Cell initiative from Janssen Pharmaceutica. The successful candidate will explore new computational methods for multi-modal integration using heterogeneous biological networks that reflect the complex biological relationships between modalities. Furthermore, the candidate will collaborate with teams focusing on the development of novel cell therapies to help realize the full potential of this approach.

Responsibilities will include:

• Development of novel machine learning methods for high-dimensional single-cell omics data aiming at the robust identification of cells types and cell states, identification of continuums of cell differentiation states and the associated gene signatures. Emphasis will be put on multiple dataset integration, time-course designs and within- and between- individuals comparative analyses ;

• Implementation of graph-representation machine learning techniques such graph embeddings and graph convolutional neural networks for supervised and unsupervised tasks on heterogeneous multiplex networks, including node, subgraph and graph classification. GNNs ML libraries such Pythorch geometric (https://pytorch-geometric.readthedocs.io/) or Deep Graph Library (https://www.dgl.ai/) will be extensively used ;

• Carrying out state-of-the-art bioinformatics analysis covering from the raw sequencing data to the functional interpretation of differentially expressed genes is expected ;

• Communication of results in the form of peer reviewed papers, seminars and communications to international conferences.

The postdoctoral position is funded for 2 years

Desired starting dates : between January and March 2024

Applicants should have a PhD in Bioinformatics, Computational Biology, Computer Science, Biostatistics, or similar, and at least one first author paper in a peer-reviewed journal or conference. Demonstrated experience in machine learning and strong data analyses skills are required. Prior experience in omics data analysis is preferable, although the laboratory will provide a thorough specific training in single-cell analysis. Knowledge/interest in functional genomics, immunology, oncology or gene therapy will be appreciated.

We are looking for knowledgeable candidates with versatile skills, with an eye on basic science and another on translational applications. Interest in the development of machine-learning algorithms, curiosity to enter into new topics and a team-building spirit are required. The position will require a constant interaction with computational, experimental labs, platforms and clinicians.

Interested candidates should send a CV and a motivation letter to antonio.rausell@inserm.fr (either in English or French)