DESeq2 results function runs very slow on Windows 10
1
0
Entering edit mode
6 weeks ago
Taha • 0

Hi,

I have bulk RNA-seq data that I analyze with DESeq2. The code I have is the following.

dds <- readRDS("DESeqDataSet.rds")
design(dds) = ~groupSSc
dds <- estimateSizeFactors(dds)
dds <- DESeq(dds)
BC_SSc <- results(dds, contrast=c("groupSSc", "SSc", "HC"),


The problem arises with the results function. When I run this code on my Mac with an M1 Pro CPU, it takes around three seconds. When I run the same code on my Windows 10 PC which has i7-12700, it takes almost a minute.

It also gives the following warning on windows 10 while on Mac, i don't get any warnings.

Warning messages:
1: In serialize(data, node$con) : 'package:stats' may not be available when loading 2: In serialize(data, node$con) :
3: In serialize(data, node$con) : 'package:stats' may not be available when loading 4: In serialize(data, node$con) :
5: In serialize(data, node$con) : 'package:stats' may not be available when loading 6: In serialize(data, node$con) :
7: In serialize(data, node$con) : 'package:stats' may not be available when loading 8: In serialize(data, node$con) :
9: In serialize(data, node$con) : 'package:stats' may not be available when loading 10: In serialize(data, node$con) :
11: In serialize(data, node$con) : 'package:stats' may not be available when loading 12: In serialize(data, node$con) :
13: In serialize(data, node$con) : 'package:stats' may not be available when loading 14: In serialize(data, node$con) :
15: In serialize(data, node$con) : 'package:stats' may not be available when loading 16: In serialize(data, node$con) :
17: In serialize(data, node$con) : 'package:stats' may not be available when loading 18: In serialize(data, node$con) :


In the end, they both gave me the same DE genes with the same results. However, there is a massive difference in performance between them.

I would appreciate it if anyone has experience with a similar problem. I am open to any suggestions. Thank you.

r deseq2 • 539 views
0
Entering edit mode

there is a massive difference in performance between them.

It still is taking only one minute.

0
Entering edit mode
6 weeks ago

Note the following:

https://support.bioconductor.org/p/115415/

it looks like it has to do with the Parallel=TRUE flag

0
Entering edit mode

This would also explain the performance difference - the code is running on multiple cores on the Apple, but only a single core on Windows.