mrcinv / gsea.py Goto Github PK

np.corrcoef is extremly slow in our use case. Write your own.

Hi, thank you so much for writing this GsEA package (it saves me time from having to implement it from scratch) 😁

Any chance you could also provide the input data used for the leukemia and pathways files below? Just so I can see how the format should be like. Thanks a lot

genes, D, C = read_expression_file("data/leukemia.txt")
gene_sets, gene_set_names = read_genesets_file("data/pathways.txt", genes)

Hey @mrcinv,

I added optional parallelization support through multiprocessing to speed up computation.
Essentially, it adds a single API change:

from gsea import gsea
# load D, C, S_sets, etc.
# ...

# runs using the existing serial computation
results = gsea(D, C, S_sets, n_jobs=1)
# runs with multiprocessing across gene sets
# using k processors
k = 2
results = gsea(D, C, S_sets, n_jobs=k)
# runs with multiprocessing across gene sets
# using all processors
results = gsea(D, C, S_sets, n_jobs=-1)

Would you be open to a PR incorporating those changes?

If so, I'm happy to make one.

Best,
Jacob