Computer Vision News - December 2021

14 NeurIPS Challenge information. We can start to look at the expression of tens of thousands of genes and hundreds of thousands to millions of cells. ” Alongside this increase in scale and resolution, there is a need for new methods to analyze the data, and a concomitant need to compare these methods and understand the relative merits of one over another. Open Problems has organized this competition to take some of the standard machine learning approaches that you can compare a whole bunch of different methods to, such as centralized benchmarking and formalized metrics, into this field of single-cell analysis, where for the most part those kinds of comparisons are performed in a one-off manner, but they want to create a living benchmark. “ One of the things we really struggle with in the field of single-cell analysis in general is understanding what our ground truth is, ” Daniel explains. “ There are different ways you can do that, but one of the things we’re taking advantage of is a new kind of technology called multimodal single-cell data . The idea is you can measure both the DNA and RNA, or both the RNA and the Daniel Burkhardt is a machine learning scientist at Cellarity, a Boston-based biotechnology company. He is also co-organizer of the Multimodal Single- Cell Data Integration competition at this year’s NeurIPS conference. He is here to tell us all about it. Daniel is part of an organization called Open Problems in Single-Cell Analysis , with Smita Krishnaswamy from Yale, Fabian Theis and Malte Lücken from Helmholtz Munich, and Angela Pisco from ChanZuckerbergBiohub. It isacollaboration that started when he was at graduate school. The goal of the organization is to standardize benchmarking in a domain of life sciences research called single-cell analysis . “ The fundamental problem we’re trying to address is an emerging datamodality called single-cell transcriptomics , or single-cell omics more broadly, ” he tells us. “ There’s been a massive influx in data over the past six years creating an incredible opportunity to look at how individual cells function. This dataexistsatascaleunheardof before inthe life sciences, particularly in terms of genetic NEURIPS CHALLENGE - MULTIMODAL SINGLE-CELL DATA INTEGRATION