Do you want to attend a seminar?

Apply for a seminar


Analysis of multiplexed assays of variant effects data with mutscan - Charlotte Soneson
Thu 13 Oct 2022 02:00 pm Institut Pasteur Auditorium Francois Jacob - BIME (26)

Multiplexed assays of variant effects (MAVE) measure the fitness of large numbers of sequence variants in a single experiment. For example, a large library of variants is created by mutating a sequence of interest (referred to as deep mutational scanning, or DMS), and the resulting pool of variants is subjected to an assay that allows for amplification or selective enrichment of sequences with desirable properties. Such desirable variants are then identified by quantifying their abundance by sequencing before and after the selection. To provide a unified, flexible interface to analyse such experiments, we developed mutscan, an R package that covers the entire workflow, from FASTQ files to count table, statistical analysis and visualisation. The core read processing module is implemented in multi-threaded C++, which enables the analysis of large sequencing experiments with reasonable time and memory constraints. Various designs of experiments are supported by mutscan, for example single or paired reads with or without unique molecular identifiers (UMIs) and flexible composition of constant and variable sequence segments in any order. In order to find variants that either increase or decrease their relative abundance upon selection, mutscan employs the widely used and well-established statistical models provided in the edgeR and limma Bioconductor packages to perform statistical analysis.

Phylogenetics with indels - Maria Anisimova
Tue 11 Oct 2022 11:00 am Institut Pasteur Retrovirus room - LWOFF (22)

Sequence alignment and phylogeny reconstruction tasks are crucial in genomics and molecular evolution. Changes between homologous characters are typically described by a Markov substitution model. In contrast, the dynamics of indels are not modelled explicitly, because the computation of the marginal likelihood even under a most simple model (e.g., TKF91) has exponential time complexity in the number of taxa. But the failure to model indel evolution may lead to artificially short alignments due to the biased indel placement, inconsistencies with phylogenetic relationships. Equally, ignoring or incorrectly treating indels has consequences for the accuracy of phylogeny reconstruction. Luckily, the classical indel model TKF91 can be modified to describe indel evolution on a phylogeny via a Poisson process, termed PIP (Bouchard-Cote, Jordan, 2012). PIP allows us to compute the joint marginal probability of an MSA and a tree in linear time. We developed a 3D dynamic programming algorithm for progressive MSA inference under PIP in polynomial time. Our MSA method is the first polynomial time progressive aligner with a rigorous mathematical formulation of indel evolution. In addition, maximum likelihood phylogeny reconstruction under the PIP model can contribute to improved accuracy, with a particularly large margin for small datasets. Consequently, we suggest how to resolve the dependency between inferences of MSAs and phylogeny by reconstructing them jointly within the same likelihood framework under the PIP model.

Mechanisms of cell plasticity in triple negative breast cancers - Celine Vallot
Thu 6 Oct 2022 02:00 pm Institut Pasteur Auditorium Francois Jacob - BIME (26)

The dynamic nature of chromatin and transcriptional features are expected to participate to tumor evolution. Our group focuses on the study of the dynamics of histone modifications in cancer cells upon cancer treatment as well as during the initial steps of tumorigenesis. We develop experimental and computational approaches to map histone marks at single-cell resolution, enabling the investigation of the dynamics of chromatin marks in tumor samples (Grosselin et al. Nat Genet 2019; Prompsy et al. Nat Comm 2020). We have recently combined single-cell epigenomic and transcriptomic approaches to lineage tracing strategies to reveal the initial epigenomic events driving tolerance to chemotherapy in triple-negative breast cancer (Marsolier & Prompsy et al., Nat Genet 2022). We show that the repressive histone mark H3K27me3 is a lock to the activation of a drug-persistent expression program in breast cancers. Under chemotherapy, very few cells can survive the treatment, and these cells have a remodeled repressive epigenome, with targeted loss at key promoters. Using demethylase inhibitor in combination to chemotherapy, we improve the response rate and delay recurrence both in vitro and in vivo. We also study mechanisms of cell plasticity in early breast tumorigenesis in vivo. We have recently mapped state transitions during Brca1-tumorigenesis in the mouse. We discovered that luminal progenitor cells undergo a partial epithelial to mesenchymal transition at the onset of tumorigenesis (Landragin &Saichi, unpublished 2022).