search

skinnider / dismay

Calculation of distance metrics for matrices
MIT License
25 stars 5 forks source link

Question about input to dismay #1

Closed kvittingseerup closed 4 years ago

kvittingseerup commented 4 years ago

Thanks for the nice package!

Is there any input type prefrence? raw counts? CPM? TPM?

(and what did you use in the recent nature methods paper?)

Cheers Kristoffer

skinnider commented 4 years ago

Hi Kristoffer, dismay is flexible about what kind of input it takes. If you provide a NxM matrix it will return a MxM matrix - so genes in columns will give you a gene-by-gene matrix, cells in columns a cell-by-cell matrix. In the paper, we used the datasets as the authors had provided them; you can find these datasets at https://github.com/skinnider/SCT-MoA. The only thing I would say is the measures that look at binary presence/absence (Hamming, Dice, Jaccard) as well as measures that specifically take zeroes into account (zero-inflated Kendall correlation) would depend on the presence of zeroes in the dataset, and so for example, normalization methods like sctransform which produce dense matrices would not be compatible with these.

kvittingseerup commented 4 years ago

Thanks for the clarification :-)