Entering edit mode
8.0 years ago
Learner
▴
280
I recently had a question related to having data as integer for DEseq .
One data looks like this will have no problem
> head(df)
treat treat1 treat2 ctrl ctrl1 ctrl2
1 95 91 88 67 65 84
2 83 97 90 88 85 89
3 99 103 99 73 83 82
4 82 99 94 82 62 72
5 88 90 96 87 61 86
6 92 101 80 74 66 62
but a data like this one will have a problem
head(df2)
treat treat1 treat2 ctrl ctrl1 ctrl2
1 0.95 0. 91 0. 88 0. 67 0. 65 0. 84
2 0. 83 0. 97 0. 90 0.88 0. 85 0. 89
3 0.99 0.103 0.99 0.73 0. 83 0.82
4 0.82 0.99 0. 94 0. 82 0. 62 0.72
5 0.88 0.90 0.96 0.87 0.61 0.86
6 0.92 0.101 0.80 0.74 0.66 0.62
Do you think if I just normalised the data to above , is OK? is there any other way to do it ?
if I don't do the normalisation, is there any other way ? or any other package ?
The question you need to ask yourself is, "regardless of how I transform my data, is the resulting analysis appropriate?" If the answer to that is, "yes!", then go ahead and "normalize" (transform) the data.
@Devon Ryan for sure, but imagine if we make integer the df2, I will get all of them zero. which kind of normalisation would you use ?
There's no generic answer to that. One first needs to know what the numbers represent. It may not be possible to convert them into a form where using DESeq2 (or equivalent) would be appropriate.