Jakwang Kim, University of British Columbia

Title: Optimal allocation in single-cell RNA sequencing: balancing between quality and quantity

Date: Friday, September 13th, 2024
Time: 1:15PM (PDT)
Location: ASB 10900

Abstract: 

 In this work, we investigate the optimal allocation rule of reads of single-cell RNA sequencing experiments. Thanks to the advance of single-cell RNA sequencing machinery, people have understood a variety of aspects of cell populations. In particular, to understand the distribution of gene expressions of cell population under the restricted budget of reads, it has been an important consideration for biologists both theoretically and practically to decide whether to inspect a few cells in detail or to explore many cells roughly. Often described as deep sequencing versus shallow sequencing, many researchers have tried to resolve this controversy and provide an optimal allocation rule of the measurements to achieve the best approximation of the true distribution of gene expression in population level under certain assumptions. Here, we establish the theory of the optimal allocation rule with full generality, which guarantees that a proposed empirical distribution is close to the true one in Wasserstein distance with high probability. Our result explicitly shows how the sparsity of gene expression and the intrinsic low dimensionality of its distribution effect on the allocation, which is the first rigorous derivation in this field. Furthermore, we propose an empirical analysis of a real data set which strongly supports our theoretical results. (based on joint work with Sharvaj Kubal and Geoffrey Schiebinger).