The upcoming CompCancer Seminar will be hosted by Eleonora Usatikova. Find the invitation below. The link is available from compcancer at charite dot de.
Dear all,
I would like to invite you to the next CompCancer seminar, which will take place on 23.02. at 3 pm.
Our guest speaker will be Benjamin Auerbach, a Ph.D. student from the Laboratory for Statistical and Translational Genomics at the University of Pennsylvania. He will present his unpublished work on circadian phase inference in single cells.
Title of the talk: “Unsupervised Circadian Phase Inference in Single-Cell RNA-Sequencing Data”
The circadian clock is a 24-hour cellular timekeeping mechanism that temporally regulates human physiology. Single-cell RNA-sequencing (scRNA-seq) has been increasingly adopted to study circadian transcription. Nonetheless, scRNA-seq has mainly been applied to data generated over light-dark cycle time courses, in which cell collection time in the light-dark cycle is presumed to be a direct proxy for cell circadian time. This assumption limits using these data for the discovery of cell phase heterogeneity, its determinants (e.g. spatial location within a tissue), and its role in other cellular processes (e.g. gating cellular differentiation). Moreover, investigators may be interested in conducting single-sample scRNA-seq of unsynchronized cell populations, for which sample collection time is not a meaningful proxy of cell phase. One approach to break the reliance on sample collection time as a proxy for cell phase is to estimate cell phase from scRNA-seq data directly, a task referred to as unsupervised phase inference. While existing approaches have been developed for similar problems, such as cell cycle phase inference in scRNA-seq and circadian phase inference from bulk RNA-seq samples, these approaches yield poor circadian phase estimates in scRNA-seq. Moreover, existing approaches do not quantify estimation uncertainty, which is essential for result interpretation from highly sparse scRNA-seq data. We’ve developed an unsupervised phase inference algorithm, Tempo, to predict cell circadian time from scRNA-seq expression. Based on Bayesian variational inference, Tempo incorporates domain knowledge of the circadian clock to yield state-of-the-art circadian phase estimates and well-calibrated uncertainty quantifications. We further demonstrate these properties generalize to the cell cycle.
On November 23rd 2021, Tincy defended her thesis “An Integrative Genetic, and Epigenetic Characterization of Pancreatic Neuroendocrine Neoplasms (PanNENs) defines Distinct Molecular Features of Endocrine- and Exocrine-like Subgroups”.