Integrate multiomics RNA and ATAC data generates trivial all-zero latent values

[elhaam]

Hi,
I am integrating single-cell RNAseq and ATAC-seq data, but iNMF generates trivial all-zero latent values for all cells of both data domains. Do you possibly know what causes this issue? I appreciate any insights on this problem.
Thanks!

[elhaam]

Just a follow-up on this: LIGER performs fine on this dataset and generates reasonable latent space for RNA and ATAc, but using iNMF, I get trivial zero latent values in 20-dimensional space. Would you please give me some tips on how I can proceed with iNMF?

Thanks!

[jw156605]

Do you mean that some cells have iNMF factor loadings that are identically 0 for all factors? — Joshua D. Welch, Ph.D. Assistant Professor Department of Computational Medicine and Bioinformatics Department of Computer Science and Engineering University of Michigan 2044 Palmer Commons, 100 Washtenaw Ave. Ann Arbor, MI 48109 Office: 734-615-0618 Administrative Assistant: Jane Wiesner<https://medicine.umich.edu/dept/dcmb/jane-wiesner> From: Elham ***@***.***> Sent: Thursday, July 21, 2022 1:21 PM To: welch-lab/liger ***@***.***> Cc: Subscribed ***@***.***> Subject: Re: [welch-lab/liger] Integrate multiomics RNA and ATAC data generates trivial all-zero latent values (Issue #265) External Email - Use Caution Just a follow-up on this: LIGER performs fine on this dataset and generates reasonable latent space for RNA and ATAc, but using iNMF, I get trivial zero latent values in 20-dimensional space. Would you please give me some tips on how I can proceed with iNMF? Thanks! — Reply to this email directly, view it on GitHub<#265 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AAS2HVLFD5X6E36JAM4KQPTVVGBGBANCNFSM53FKEE7Q>. You are receiving this because you are subscribed to this thread.Message ID: ***@***.***> ********************************************************** Electronic Mail is not secure, may not be read every day, and should not be used for urgent or sensitive issues

[elhaam]

Hi @jw156605 thank you for your response. Yes, that is correct.