search

MarioniLab / MNN2017

Code for the MNN manuscript figures
51 stars 19 forks source link

Runtime #11

Closed chenlingantelope closed 5 years ago

chenlingantelope commented 5 years ago

The following dataset contains ~76K cells and according to the paper should take ~20mins but I ran it for >24h and obtained no output. Both of them are from the 10x website https://support.10xgenomics.com/single-cell-gene-expression/datasets/2.1.0/pbmc8k https://support.10xgenomics.com/single-cell-gene-expression/datasets/1.1.0/fresh_68k_pbmc_donor_a

data1 = 'data/pairwise/pbmc8k.mtx' data2 = 'data/pairwise/pbmc68k.mtx' data <- lapply(c(data1,data2),function(fname){ X = readMM(fname) return(t(X)) }) library(scran) ibrary('gmodels') data1= as.matrix(data[[1]]) data2= as.matrix(data[[2]]) start = proc.time() mnn_correct <- mnnCorrect(data1, data2) # corrected values end = proc.time()

LTLA commented 5 years ago

If you read the README, it'll direct you to https://github.com/MarioniLab/FurtherMNN2018.

But to answer your immediate question, I would suggest some feature selection before throwing it into mnnCorrect. It's not like removeBatchEffect, it doesn't operate on a gene-by-gene basis.