Study: Columbia University/NYP COVID-19 Lung Atlas 116313 cells

A molecular single-cell lung atlas of lethal COVID-19


COVID-19 Autopsy Study Summary

Understanding cellular drivers of lethal SARS-CoV-2 infection is an essential first step towards improving therapies for severe COVID-19. We performed single-nuclei RNA-seq (snRNA-seq) on >116,000 nuclei from 19 COVID-19 autopsy lungs and seven pre-pandemic controls. Lungs from COVID-19 decedents show distinct fractional and (dys)functional changes across the immune and non-immune cellular landscape. This data set provides a rich single-cell landscape of lethal COVID-19 and initial insights into the cellular interplay that shapes the highly inflamed and remodeled lung ecosystem. The lung atlas presented here is the first in a series of data releases covering different organs profiled from COVID-19 patients.


Lung Study Summary

Here, we share an early release of 116,314 snRNA-seq profiles across 20 frozen lungs obtained from 19 COVID-19 decedents and seven control patients with short postmortem interval (PMI) autopsies. The COVID-19 cohort comprises seven female and 12 male decedents, including 13 patients of Hispanic ethnicity, with an age range from 58 to >89 years who had acquired SARS-CoV-2 infection and succumbed to the disease. The average time from symptom onset to death was 27.5 days (range, 4–63 days). After rapid autopsy with a median PMI of 4 hours (range 2–9 hours) collected tissues were either flash-frozen or frozen following OCT (optimal cutting temperature) embedment and subjected to snRNA-seq using a droplet-based platform (10x Genomics). All included patients had underlying hypertensive disorder and frequently one or more additional co-morbidities associated with increased risk for severe COVID-19.

Single-cell sequencing data was demultiplexed into FASTQ files and aligned to a custom-built joint human and SARS-CoV-2 genome using Cell Ranger (v.5.0.0). We used CellBender (v.0.2.0) to remove technical artifacts and ambient RNA counts from the gene expression matrices. The resulting individual count matrices underwent quality control and were integrated using Seurat (v.3.2.3). To gain an overview of the cellular landscape of affected lungs, we identified different cell types iteratively on three levels of granularity (main, intermediate, and fine) using a three-pronged approach for cell type identification: unbiased identification of cluster markers, discovery of cell types using signatures from reported atlases, and manual curation to sub-stratify cell populations and cell states (that may be unique to COVID-19) using expert knowledge.

We provide the gene expression matrix, cell clustering, UMAP dimensionality reduction coordinates, cell type assignments, and associated metadata on the Single Cell Portal. In parallel, another SARS-CoV-2 lung atlas, generated at the Broad Institute, is being released at



A molecular single-cell lung atlas of lethal COVID-19. Nature (2021)



Benjamin Izar, MD, PhD; 650 West 168 Street; William Black Building, 1708E; 10032, New York, NY;



Johannes C. Melms, Jana Biermann, Huachao Huang, Yiping Wang, Ajay Nair, Somnath Tagore, Igor Katsyv, André F. Rendeiro, Amit Dipak Amin, Denis Schapiro, Chris J. Frangieh, Adrienne M. Luoma, Aveline Filliol, Yinshan Fang, Hiranmayi Ravichandran, Mariano G. Clausi, George A. Alba, Meri Rogava, Sean W. Chen, Patricia Ho, Daniel T. Montoro, Adam E. Kornberg, Arnold S. Han, Mathieu F. Bakhoum, Niroshana Anandasabapathy, Mayte Suárez-Fariñas, Samuel F. Bakhoum, Yaron Bram, Alain Borczuk, Xinzheng V. Guo, Jay H. Lefkowitch, Charles Marboe, Stephen M. Lagana, Armando Del Portillo, Emmanuel Zorn, Glen S. Markowitz, Robert F. Schwabe, Robert E. Schwartz, Olivier Elemento, Anjali Saqi, Hanina Hibshoosh, Jianwen  Que, Benjamin Izar

Corresponding authors

Related publications
A molecular single-cell lung atlas of lethal COVID-19
Nature volume 595, pages114–119 (2021)