Trying PHLOWER for general sample x feature table

[liezeltamon]

Hi @chengmingbo,

Thanks for this tool!

I just came across this so I would have to read the paper first but I thought I should ask if you think I could use PHLOWER for non-single-cell data and if yes, are there certain parts that I should be wary of.

So my dataset is single-cell transcriptome, proteome, VDJ data and from this I have calculated different features per patient sample (n = 130). Features include mean expression per cell type, other types of pathway activity scores, VDJ gene usages and other repertoire data. But it would be great if I can use the tool for any feature table I have in the future. So my feature table is 130 patient samples x thousands of features.

This is the expected trajectory based on the patient groups:

Groups are imbalanced with some samples having 20 and some just 5 patient samples.

Thank you for the help!

Best,
Liezel

[chengmingbo]

Hi Liezel,
It’s an honor that you’re interested in using PHLOWER for patient-level analysis. If each patient can be treated as a “single cell,” the method might be applicable, although I haven’t tested PHLOWER on this type of data, so I can’t be certain about the results.

An alternative I would suggest is PILOT (https://pilot.readthedocs.io/en/latest/), which was also developed in our lab and may be more suitable for patient-level analysis.

[liezeltamon]

Thanks will have go!