Research Projects

Modelling TCR repertoire and specificity

T cells have the ability to generate billions of T-Cell Receptors that can recognize various epitopes displayed on HLA molecules. While information about the presence of specific T cells in a tumor can be obtained with TCR-sequencing technologies, a major challenge remains to know which T cells recognize which epitopes. To address this question we are combining machine learning algorithms with multiple experimental strategies to screen TCRs recognizing different epitopes and train predictors of TCR-epitope interactions. In [Croce et al. 2024], we describe the first version of our tools (MixTCRpred) and how it can be used to analyse both bulk and epitope-specific TCR repertoires.

Related tools

Analysis and predictions of antigen presentation and TCR recognition

The diversity of T-cell epitopes in cancer is overwhelming due the heterogeneity of genetic alterations and the polymorphism of MHC genes. To narrow down the most promising candidates, our lab has developed state-of-the-art predictors of MHC-I [Gfeller et al. Cell Systems 2023] and MHC-II ligands [Racle et al. Nature Biotech 2019; Immunity 2023], as well as predictors of neo-epitope TCR recognition [Schmidt et al. Cell Rep Med 2021; Gfeller et al. Cell Systems 2023]. These predictions are largely based on high-quality MHC peptidomics data and machine learning algorithms for motif deconvolution [Bassani-Sternberg and Gfeller J Immunol 2016, Racle et al. Nature Biotech 2019]. Our findings provide detailed MHC motif [Tadros et al. NAR 2023] and revealed novel structural properties of MHC molecules [Guillaume et al. PNAS 2018; Racle et al. Immunity 2023].

Related tools

Bulk and single-cell genomics analyses of tumors

Tumors are composed of heterogeneous cell types, comprising both cancer cells and non-maligant cells. The presence and phenotype of these different cell types plays an important role in tumor progression and response to therapy. Our lab has developed computational tools to simultaneously Estimate the Proportion of Immune and Cancer cells (EPIC) from bulk tumor RNA-seq and ATAC-seq data that can quantitatively predict the fraction of all major immune cell types, as well as cancer cells [Racle et al. 2017; Racle et al. 2020; Gabriel et al. 2023]. In parallel, we have developed a powerful approach to facilitate the analysis of large single-cell genomics data based on the concept of ‘metacells‘ (read more about this in [Bilous et al. 2022], and our recent review/tutorial [Bilous et al 2024]).

Related tools

Selected Publications

Machine learning predictions of MHC-II specificities reveal alternative binding mode of class II epitopes. Racle J, Guillaume P, Schmidt J, Michaux J, Larabi A, Lau K, Perez MAS, Croce G, Genolet R, Coukos G, Zoete V, Pojer F, Bassani-Sternberg M, Harari A, Gfeller D. (2023) Immunity, 2023, 56, 1-17.

Deep learning predictions of TCR-epitope interactions reveal epitope-specific chains in dual alpha T cells. Croce G, Bobisse S, Moreno DL, Schmidt J, Guillaume P, Harari A, Gfeller D. (2024) Nature Communications, 5(1):3211.

Metacells untangle large and complex single-cell transcriptome networks. Bilous M, Tran L, Cianciaruso C, Gabriel A, Michel H, Carmona, S, Pittet MJ, Gfeller D., 2022. BMC Bioinformatics, 23, 336.

Open Positions

Currently, we do not have open positions. But if you are still interested in working in my group, contact me directly and we will see if there is a match. We also strongly encourage and provide support for candidates interested in applying to their own PhD/post-doc fellowships.