How to do a DESq analysis using a table with raw counts or normalized counts

Question

[Lila M]

Hello community!

I have a , maybe, silly question. I have downloaded table with raw counts and normalized counts that looks like this

sample_id    raw_read_count   gene_id       normalized_read_count 
168-2-1      9685          ENSG00000000003   48.56 
168-2-1      25            ENSG00000000005   0.25 
168-2-1      4630          ENSG00000000419   47.89 
168-2-1      1847          ENSG00000000457   6.81

and other table with the metadata info

  sample_id  treatment
   168-2-1    Pre-treatment
   004-1-5    Pre-treatment
   005-1-8    Pre-treatment
   034-1-0    Pre-treatment
   034-3-8    Post-treatment

And I was wondering how is the best way to do a DE analysis using data, as I always start from RNAseq samples, I don't have much idea about how to start. Any advice or tip would be great!! Thanks!!

Answer 1

[rpolicastro]

You want to convert your data into a wide format raw count matrix first.

Your example counts data.

cts <- structure(list(sample_id = c("168-2-1", "168-2-1", "168-2-1", 
"168-2-1"), raw_read_count = c(9685L, 25L, 4630L, 1847L), gene_id = c("ENSG00000000003", 
"ENSG00000000005", "ENSG00000000419", "ENSG00000000457"), normalized_read_count = c(48.56, 
0.25, 47.89, 6.81)), class = "data.frame", row.names = c(NA, 
-4L))

Using tidyverse for the convenience.

library("tidyverse")

mat <- cts %>%
  select(!normalized_read_count) %>%
  pivot_wider(names_from=sample_id, values_from=raw_read_count) %>%
  column_to_rownames("gene_id") %>%
  as.matrix

> mat
                168-2-1
ENSG00000000003    9685
ENSG00000000005      25
ENSG00000000419    4630
ENSG00000000457    1847

This matrix and the meta-data file can then be used as input to DESeq2

[Lila M]

Great solution, Thank you very much