Support of gz / npz format as input

[LeilyR]

Would be nice if the gz can be used as input, txt files can get so big. I have a matrix on fly which is 14G in txt as opposed to 100M in gz. It is not even really feasible to upload it to galaxy

[gtrichard]

Yes I agree, this is planed for next release :)

[gtrichard]

Should be working with last version.

[LeilyR]

which version support that? 0.4? The help message still says: DeepTools file obtained from computeMatrix --outFileNameMatrix and I get this error :
2020-05-18 16:37:05 Extracting metadata from file:
matrix.gz ... Done!
Warning messages:
1: NAs introduced by coercion
2: NAs introduced by coercion
Error in metadata$scores[[i]] : subscript out of bounds
Execution halted

command line is:
dsCompareCurves -i matrix.gz -o deepstats_on_matrix -p 20 -c 'regions'

Note: matrix is fine and has been already visualised by both plotHeatmap and plotProfile. It has one score file and 2 regions.

[gtrichard]

Hi Leily,

Yes Katarzyna also posted that issue, I need to change the metadata parser to fully support npz format. After checking back what I did a while ago, I took care of the zipped part of the problem, not about how the metadata / header of the files is looking. And indeed the metadata are stored in a very different way in npz vs txt format.

During the meantime, please use the outFileNameMatrix output as input for dsCompareCurves.

[LeilyR]

Thanks, I did use the text file and it works :) . Just keep in mind that computeMatrix output format is not npz but gz.