plotSignatures issue about maftools

[PoisonAlien]

Hi,
Is the maf used here is from package (LAML)? I see matrix of dim 87x96, i.e 87 samples. Example maf has 196 samples. Did you subset it ?

Regardless, I should admit that the way comparison made to known signature is very naive here and can go off-set sometimes. Results printed are based on maximum simple correlation coefficient. Better way to make a comparison is using cosine similarity. (Sorry I will try to implement it into next version).

For now you can draw correlation matrix of output and see which signature makes more sense than the printed one. (comparison are made to these signatures

pheatmap::pheatmap(laml.sign$corTable, cluster_cols = F, cluster_rows = F)

[shirleydaidai]

Thanks for you quick response.
Another question I just noticed, why you use only 21 validated signatures not 30 on cosmic website?
The last few signatures might be useful to my research.

[PoisonAlien]

Yes you are correct. I wanted to use those updated 30 signatures mentioned on cosmic, but the raw data is not available for them (I couldn't find online, maybe its out there somewhere). For the original 21 signatures, results are downloaded from here for comparison. You can also find the same in extdata of maftools installation directory.

[shirleydaidai]

I look forward to the new version.
This tool is very helpful to me as coding naive person.
Many thanks.