Log2 raw count matrix with one condition

Question

[User000]

Hello, I have a gene count matrix with 1 condition only. I have made log2(raw_count+1) with DESeq2, but when I filter off genes with zero expression my gene names disappear. How to fix this so that at the end I have my gene names and log2 values?

raw_count <- read.csv("gene_count_matrix.csv", header = TRUE, row.names = 1)
# Remove all gene which has 0 value in all sample
all <- apply(raw_count, 1, function(x) all(x==0))
newdata <- raw_count[!all,]
write.csv(newdata, file = "filtered_matrix.csv")
final_count <- read.csv("filtered_matrix.csv", header=TRUE, row.names = 1)

Answer 1

[Carlo Yague]

You should use drop=FALSE to prevent droping the dimensions of the matrix.

newdata <- raw_count[!all, , drop=F]

[User000]

this works, thank you! strangely, when I have many conditions this does not happen, do you know why? thank you again

[Carlo Yague]

Because if you have only one column in your matrix/dataframe, the default behaviour of R is to convert it into a vector. Edit : more info in the R help :

?'['

Answer 2

[russhh]

I can't reproduce your issue:

M <- matrix(c(0,0,0,1,0,1, 1, 2,3 ), nrow = 3, byrow = TRUE)
rownames(M) <- letters[1:3]
drop.rows <- apply(M, 1, function(x) all(x==0))  # best not to use the function-name `all`
M[!drop.rows, ]

> M[which(!drop.rows), ]
  [,1] [,2] [,3] 
b    1    0    1   
c    1    2    3

Could you post a minimal example, please.

[Carlo Yague]

Try with only 1 column :

M <- matrix(c(0,0,0,1,0,1, 1, 2,3 ), nrow = 9, byrow = TRUE)
rownames(M) <- letters[1:9]
drop.rows <- apply(M, 1, function(x) all(x==0))  # best not to use the function-name `all`

M[!drop.rows, , drop=T]
d f g h i 
1 1 1 2 3 

M[!drop.rows, , drop=F]
  [,1]
d    1
f    1
g    1
h    2
i    3