search

dylkot / cNMF

Code and example data for running Consensus Non-negative Matrix Factorization on single-cell RNA-Seq data
MIT License
243 stars 57 forks source link

Question about negative value in gep_scores matrix #64

Closed Tang-RH closed 1 month ago

Tang-RH commented 1 year ago

Dear authors: Thank you for your work on developing the nice tool cNMF. When I run the PBMC tutorial https://github.com/dylkot/cNMF/blob/master/Tutorials/analyze_pbmc_example_data.ipynb, I found that the gep_score ,the result matrix to select top genes in a GEP, had some negative values. And I saw the similar result in my own data. image I want to confirm that if the gep_score matrix is one of the nmf result matrixes. If so, why there are some negative values in it. If not, what is it, and why can we use it to select the top genes in a GEP?

dylkot commented 1 month ago

Hi, gep_score is actually not produced directly by NMF. It is produced by performing ordinary least square regression to predict the each gene but in mean-centered, standard deviation normalized units. So it is expected to have negative values.