sijin-zhanglab / panmyeloid Goto Github PK

Hi Sijin,

Hope you are doing well and don't mind me asking some questions about the methods or workflows regarding the scRNAseq analysis part.

I noticed in the paper, you have done some integration (human cell landscape)HCL dataset with your own 10X dataset by Scanorama, and used the integrated corrected expression value for comparison on some genes.
We want to do something similar but have failed on the integration. It seems like almost every gene shows same trend that HCL cells have higher expression level compared to the 10X.
So I am wondering could you share some of your insights or suggestions on how to do the integration via Scanorama?

Thanks!
Any response would be appreciated!
Shaowen

Hi Sijin,

Nice job and Congrats. I hope you don't mind if I ask some questions about the analytical procedure.

1). As is mentioned in Results, you and your team analyzed each dataset independently and performed marker-based annotation. But when I tried to reproduce the result using LUNG, MEL and newly generated datasets, I can hardly see the enrichment of the cell markers you provided In the paper. (Figure below is from LUNG dataset and I followed the protocol from the paper with same pacgage version) Could you give me some advice about how to annotate the dataset on such situation?

2). I also noticed that MEL datasets used in the paper only available for TPM data, so should it be normalized by library-size further? (normalize_total() and loglp() were employed) Would it cause something like 'over-normalization'?

Thanks for your time and all the best,

Yiling