How To Input Missing Values For A Marker In Dexseq

Question

[venks]

Hi,

I have 100 samples (individuals) and 100 (SNPs) genotyped. I call the SNPs either a WildType (WT) or a Rare-variant (RV) across all samples. Few SNPs in a sample had very poor quality. So It is a missing value. I read in this case/control information as matrix. How do we read the missing value in DEXSeq analysis ??? Should I set them to '0' or just leave it blank?

[Devon Ryan]

Could you maybe give some more information? Are you looking for the effect of the rare variants on differential exon usage? Are you suspecting that the genes harboring (or near) the variants will show a difference, or are the variants in a gene that affects splicing?

[venks]

I am looking for the effect of rare variant. My question is very straightforward. What do DEXSEQ expect to see when one of my marker/SNP is not called for one particular sample (because we are not sure if it is genotyped right for a certain sample) but is called a RV or WT for all other samples. Does DEXSEQ expect a "BLANK SPACE" or <na> or 0

[Devon Ryan]

Sorry, but your reply isn't helpful. Are you trying to use a design matrix with 100 factors (wherein the appropriate sample->factor assignment is occasionally unknown)? That's the most intelligible thing I can make from what you've written. If that's correct, then please edit your question above to make this explicit. If that's not what you're trying to do, then edit your question to EXPLICITLY state EXACTLY what you're trying to do ("looking for the effect of rare variant" is largely meaningless).

[venks]

I am sorry. Actually I am trying to find the differential exon usage having beadxpress genotyping data. But I guess it is not relevant to my question. Again my question is how do we input a missing case/control information in DEXSEQ. But I guess I figured it out. In DESEQ we give 0 for a no call.. Since DEXSEQ is from the same developers I guess DEXSEQ also requires 0.

Thanks again for your effort to help @dpryan.

[Nicolas Rosewick]

DEXSeq use read count from RNA-Seq data. Do you have such data ? otherwise DEXSeq is not suitable.

[venks]

Yes I have the count data. I was just wondering when I input the case/control information. For eg: SNP123 in sample1 is not called (because of some genotyping error) but the same SNP is called WT/RV for all other samples. How do I input the missing value? Should I leave it as <na> (or) "a blank space" (or) 0. Hope my question makes sense.

Thanks

Answer 1

[venks]

You can input the missing values as 0. But we have to make sure we have specified the conditions in the design else 0 is also considered as a condition variable.