Thread by @thatdnaguy

#ASHG21 ALP: Alkes L Price.
Identifying disease-critical cell types and cellular processes across the human body by integration of single-cell profiles and human genetics.
Preprint - biorxiv.org/content/10.110…

#ASHG21 ALP: Cellular dysfunction leads to disease. We can use GWAS data to find the critical cell types.

#ASHG21 ALP: They wanted to use scRNA-Seq to help understand GWAS. Use the sc-linker program. [ I see no GitHub link ]

#ASHG21 ALP: cell type gene programs are genes that are specifically expressed in annotated cell types using known marker genes. Examples include LYZ in monocytes and IL32 in T cells

#ASHG21 ALP: sanity check to link blood cell types to blood cell traits. Confirm expectation (like blood cell types linked to traits of lymphocyte percent, rbc count, etc)

#ASHG21 ALP: immune-related gwas show signals in immune cells, as expected. brain related traits split into gabanergic and glutamertergic cell types.

#ASHG21 ALP: Important to use tissue specific cell types to link to traits that would be in that tissue.

#ASHG21 ALP: compare sc-linker to MAGMA. Generally saw greater significance with sc-linker.

#ASHG21 ALP: what about cell processes? instead look at processes that may exist across different cell types. Use unsupervised negative matrix factorization (NMF).

#ASHG21 ALP: Can also look at disease progression programs, i.e. genes specifically expressed in disease samples compared to healthy samples in the same cell type.

#ASHG21 ALP: Example - enterocytes and M cells show disease progression association with ulcerative colitis.

#ASHG21 ALP: Can zero in on individual genes driving disease progression signals. Prioritize genes based on grade in the gene program and MAGMA gene score. Example - atrial fibrillation in cardiomyocytes finds KD2L2.