Differential Expression Analysis : Question regarding DEseq2 Input

Question

[venura]

Hi,

I want to do a differential expression analysis using DESeq2 and I have a question regarding the input. Currently I have csv file with following information for each sample. Unfortunately, I dont have access to raw data. I am trying to figure out how to input this data. Can I straightaway use expected counts?

gene_id transcript_id(s)    length  expected_count  FPKM
AT1G01010   AT1G01010.1 1688    79  2.43

Can some one help me to go forward with my analysis? Thank you in advance. VH

[igor]

That looks like RSEM output. You can process that for DESeq2 analysis with tximport.

[venura]

Thank you so much. I will use tximport. However, there is a problem compared to the example given there. I have individual files with count data. They have different number of rows.

• Is there a way to input files separately or a quick way combine them based on gene names?
• What should I do about genes missing genes (those didnt appear since there there are no counts)? Thank you!

[Kevin Blighe]

I don't believe tximport supports different numbers of genes in your samples; so, you could try:

1. Read all samples into a list object
2. Derive a unique list of gene names across all samples
3. Using the unique list, for each sample, add extra rows (where needed) for missing genes. Fill these rows with 0 / zeros

[igor]

I would be concerned about different number of rows. That suggests that the samples were quantified against different versions of the transcriptome, so there may be differences simply due to the same being gene not represented by the same sequence.

[venura]

I guess I will somehow try get raw files and do the analysis from scratch!