Question in applying DESeq2
1
0
Entering edit mode
krisd3v • 0
@krisd3v-15917
Last seen 5.8 years ago

Hi there,

I'm one of the users of DESeq2. Recently I got a bunch of data with only raw read counts of RNA-seq. The design was quite weird. They applied the same procedure to the replicates but got sequenced in completely different platforms(Solid, Nextseq and Hiseq). If the procedures from fastq files to raw counts are the same, can I simply apply them into DESeq2 and find the results of differential genes?

Thanks.

deseq2 • 480 views
ADD COMMENT
0
Entering edit mode
@mikelove
Last seen 3 hours ago
United States

Can you type out the "colData" for this experiment, that is, a table where the rows are samples and the columns are variables, e.g. a condition taking values "control" or "treatment", as well as platform.

ADD COMMENT
0
Entering edit mode

Thanks for the reply.

In that case, the case would be rows with all genes expressed founded in RNA-seq, the colname would be "untreated1","untreated2","untreated3","treated1","treated2","treated3", platform would be "untreated1" & "treated1"  SOLiD Total RNA-Seq kit  on the SOLiD System at 50M reads, "untreated2" & "treated2" RNA libraries prepared with the Ultra Directional RNA Library Prep Kit for Illumina on the Illumina NextSeq System at 50M reads, "untreated3" & "treated3" Ultra Directional RNA Library Prep Kit for Illumina on the Illumina HiSeq 4000 PE100 System at 50M reads. (These 3 untreated and treated pairs were biological replicates, which I think sequencing on different platform was a mistake in research design. I don't know whether my design on merging the data together works in this case.)

Thanks.

ADD REPLY
1
Entering edit mode

Here is closer what I was asking for, e.g. a sample table which forms the 'colData' of a Bioconductor object:

condition platform
untreated SOLiD
untreated NextSeq
untreated HiSeq
treated   SOLiD
treated   NextSeq
treated   HiSeq

You can use a design of ~platform + condition, and then test on the condition variable. Simply calling DESeq() followed by results() will work if 'condition' is the last variable in the design, but you should make 'untreated' the reference level of 'condition' (see note on factor levels in the vignette). This will control for the differences due to platform in this dataset by fitting a baseline for each platform.

ADD REPLY
0
Entering edit mode

Thanks, Michael.

I've proceeded that and got very good results.

ADD REPLY

Login before adding your answer.

Traffic: 798 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6