search

JiekaiLab / scTE

MIT License
87 stars 27 forks source link

Not getting significant read counts #51

Open A1aNNN opened 1 year ago

A1aNNN commented 1 year ago

Hi there,

I've been trying to run scTE on a dataset of human PBMCs. I ran CellRanger count on the FASTQs, got a bam file, created a human genome index with scTE_build using the same genes.gtf file that was used in the CellRanger reference package, then ran scTE on that bam file.

The results from scTE said that there were 1172 cells detected expressing at least 200 genes, and the results from running CellRanger count also indicated that there were an estimated 3500 genes expressed per cell. However, the csv file had almost no reads for the genes. Only 4 genes were found to have read counts of over 100 within some cells, which doesn't seem like it would be enough to separate the cells into clusters.

Attached below are some screenshots illustrating what I ran, the CellRanger count output, as well as a screenshot of my csv file. Just wondering if you had any input on this issue. Thank you in advance!

Screen Shot 2022-12-13 at 10 14 14 AM Screen Shot 2022-12-12 at 12 07 45 PM Screen Shot 2022-12-12 at 11 59 35 AM

metrics_summary_new.xlsx

jphe commented 1 year ago

It's wired, as scTE reported there were 1172 cells detected expressing at least 200 genes, so the outputed csv file should have at least 200 genes with counts >=1, can you confirm this?

Only 4 genes were found to have read counts of over 100 within some cells, if I am right, there should be many genes (at least 200) expression <100 but >=1. Can you figure up the Median Genes per Cell for the genes with >=1 counts as the Cell Ranger for the scTE outputed csv table.

A1aNNN commented 1 year ago

Hi, thanks for the reply!

Yes, there are genes with read counts >=1 so that is what scTE is reporting when it says that there were 1172 cells found expressing at least 200 genes. The bam file I inputted in was after running CellRanger count on some 10x Genomics test data of ~1000 PBMCs, which I know should have many differently expressed genes.

I may have just ran CellRanger count wrong so I'll try fixing that up and update this issue once I get a new bam file to test scTE on.

swainasish commented 4 months ago

As you are using Cellranger: use 

-CB CB -UMI UB

Note: Developers can confirm this.