laurahspencer/DuMOAR

Perform DESeq2 analysis on RNASeq data

Opened this issue · 3 comments

laurahspencer commented
Perform DESeq2 analysis on RNASeq data
laurahspencer commented

Ran DESeq2 analysis on gene count matrix that Giles produced using STAR and all RNASeq data aligned to genome (not scrubbed of non-arthropoda reads). Here's my RMarkdown Notebook, RNASeq.html. A quick summary:

I removed quite a few low-frequency genes:

  • No. genes before filtering: 43,800
  • No. genes remaining after pre-filtering: 7,075
  • No. of genes dropped: 36,725
  • % of fragments remaining after pre-filtering: 95.551%

Below is a PCA of the two sign. principal components. A perMANOVA test indicates the treatments vary in multivariate space. There's also a heat map of the top 10% most variable genes (~700 genes). Differential expression analysis found a total of 40 DEGs. No significantly enriched biological processes. I suspect that improving our alignment will increase the no. of DEGs.

image

image

laurahspencer commented

Here is a CSV file with DEG stats and annotation info. Of the 40 DEGs 11 were annotated: DEGs.csv

Here's gene names for those that are annotated (from DAVID):
image

laurahspencer commented

NEW analysis approach:

  • @ggoetznoaa align all reads (NOT megan-scrubbed) to "original" trinity de-novo transcriptome using ALL data (NOT megan-scrubbed reads)
  • get @laurahspencer the gene count matrix
  • @laurahspencer blast the "original" transcriptome to Uniprot database
  • @laurahspencer run the gene count matrix through DEG analysis