Distribution shifts -- where the training distribution differs from the test distribution -- can substantially degrade the accuracy of machine learning (ML) systems deployed in the wild. Despite their ubiquity, these real-world distribution shifts are under-represented in the datasets widely used in the ML community today. To address this gap, we present WILDS, a curated collection of 8 benchmark datasets that reflect a diverse range of distribution shifts which naturally arise in real-world applications, such as shifts across hospitals for tumor identification; across camera traps for wildlife monitoring; and across time and location in satellite imaging and poverty mapping. On each dataset, we show that standard training results in substantially lower out-of-distribution than in-distribution performance, and that this gap remains even with models trained by existing methods for handling distribution shifts. This underscores the need for new training methods that produce models which are more robust to the types of distribution shifts that arise in practice. To facilitate method development, we provide an open-source package that automates dataset loading, contains default model architectures and hyperparameters, and standardizes evaluations. Code and leaderboards are available at this https URL.

WILDS: A Benchmark of in-the-Wild Distribution Shifts

The brain, with its diverse physiology and intricate cellular organization, is the most complex organ of the mammalian body. To expand our basic understanding of the neurobiology of the brain and its diseases, we performed a comprehensive molecular dissection of 10 major brain regions and multiple subregions using a variety of transcriptomics methods and antibody-based mapping. This analysis was carried out in the human, pig, and mouse brain to allow the identification of regional expression profiles, as well as to study similarities and differences in expression levels between the three species. The resulting data have been made available in an open-access Brain Atlas resource, part of the Human Protein Atlas, to allow exploration and comparison of the expression of individual protein-coding genes in various parts of the mammalian brain.

An atlas of the protein-coding genes in the human, pig, and mouse brain

A primary goal of human genetics is to identify DNA sequence variants that influence biomedical traits, particularly those related to the onset and progression of human disease. Over the past 25 years, progress in realizing this objective has been transformed by advances in technology, foundational genomic resources and analytical tools, and by access to vast amounts of genotype and phenotype data. Genetic discoveries have substantially improved our understanding of the mechanisms responsible for many rare and common diseases and driven development of novel preventative and therapeutic strategies. Medical innovation will increasingly focus on delivering care tailored to individual patterns of genetic predisposition.

/pdf/a-brief-history-of-human-disease-genetics-s7ecc5yuyc.pdf

A brief history of human disease genetics.

Rheumatoid arthritis (RA) is a chronic immune-mediated disease that primarily affects the synovium of diarthrodial joints. During the course of RA, the synovium transforms into a hyperplastic invasive tissue that causes destruction of cartilage and bone. Fibroblast-like synoviocytes (FLS), which form the lining of the joint, are epigenetically imprinted with an aggressive phenotype in RA and have an important role in these pathological processes. In addition to producing the extracellular matrix and joint lubricants, FLS in RA produce pathogenic mediators such as cytokines and proteases that contribute to disease pathogenesis and perpetuation. The development of multi-omics integrative analyses have enabled new ways to dissect the mechanisms that imprint FLS, have helped to identify potential FLS subsets with distinct functions and have identified differences in FLS phenotypes between joints in individual patients. This Review provides an overview of advances in understanding of FLS biology and highlights omics approaches and studies that hold promise for identifying future therapeutic targets.

Restoring synovial homeostasis in rheumatoid arthritis by targeting fibroblast-like synoviocytes

The Human Tumor Atlas Network: Charting Tumor Transitions Across Space and Time at Single-Cell Resolution

Author(s): Snyder, Michael P; Lin, Shin; Posgai, Amanda; Atkinson, Mark; Regev, Aviv; Rood, Jennifer; Rozenblatt-Rosen, Orit; Gaffney, Leslie; Hupalowska, Anna; Satija, Rahul; Gehlenborg, Nils; Shendure, Jay; Laskin, Julia; Harbury, Pehr; Nystrom, Nicholas A; Silverstein, Jonathan C; Bar-Joseph, Ziv; Zhang, Kun; Borner, Katy; Lin, Yiing; Conroy, Richard; Procaccini, Dena; Roy, Ananda L; Pillai, Ajay; Brown, Marishka; Galis, Zorina S; Cai, Long; Shendure, Jay; Trapnell, Cole; Lin, Shin; Jackson, Dana; Snyder, Michael P; Nolan, Garry; Greenleaf, William James; Lin, Yiing; Plevritis, Sylvia; Ahadi, Sara; Nevins, Stephanie A; Lee, Hayan; Schuerch, Christian Martijn; Black, Sarah; Venkataraaman, Vishal Gautham; Esplin, Ed; Horning, Aaron; Bahmani, Amir; Zhang, Kun; Sun, Xin; Jain, Sanjay; Hagood, James; Pryhuber, Gloria; Kharchenko, Peter; Atkinson, Mark; Bodenmiller, Bernd; Brusko, Todd; Clare-Salzler, Michael; Nick, Harry; Otto, Kevin; Posgai, Amanda; Wasserfall, Clive; Jorgensen, Marda; Brusko, Maigan; Maffioletti, Sergio; Caprioli, Richard M; Spraggins, Jeffrey M; Gutierrez, Danielle; Patterson, Nathan Heath; Neumann, Elizabeth K; Harris, Raymond; deCaestecker, Mark; Fogo, Agnes B; van de Plas, Raf; Lau, Ken; Cai, Long; Yuan, Guo-Cheng; Zhu, Qian; Dries, Ruben; Yin, Peng; Saka, Sinem K; Kishi, Jocelyn Y; Wang, Yu; Goldaracena, Isabel; Laskin, Julia; Ye, DongHye; Burnum-Johnson, Kristin E; Piehowski, Paul D | Abstract: Transformative technologies are enabling the construction of three dimensional (3D) maps of tissues with unprecedented spatial and molecular resolution. Over the next seven years, the NIH Common Fund Human Biomolecular Atlas Program (HuBMAP) intends to develop a widely accessible framework for comprehensively mapping the human body at single-cell resolution by supporting technology development, data acquisition, and detailed spatial mapping. HuBMAP will integrate its efforts with other funding agencies, programs, consortia, and the biomedical research community at large towards the shared vision of a comprehensive, accessible 3D molecular and cellular atlas of the human body, in health and various disease settings.

/pdf/the-human-body-at-cellular-resolution-the-nih-human-24oc8mpble.pdf

Amanda Posgai

Papers

The human body at cellular resolution: the NIH Human Biomolecular Atlas Program