hemberg-lab / d3e Goto Github PK

There seems to be an error in the wiki for making the control file, as well as in the file name for the .py script.

A control file can be created by using makeControl.py which has the following syntax:

python makeControl inputFile outputFile label

Hence in our case we run:

python makeControl example.txt example_control.txt G1

Should be:

python D3EMakeControl.py inputFile outputFile label

And the filename has a typo, it's missing the 'r' in Control.

Thanks!

I noticed D3EMergeOutput.py removes rows -- always the first row after the header.
I think line 37 in D3EMergeOutput.py should be commented out.

I have a problem concerning p values in the output table of D3E.
A lot of them are NaN (around 80%).

I run the analysis in this sequence:

D3EMakeContol.py infile infile_ctrl Group1

D3ECmd.py infile infile.out Group1 Group2 -m 1 -t 0 -z 0 -n 0
D3ECmd.py infile_ctrl infile_ctrl.out Group1_1 Group1_2 -m 1 -t 0 -z 0 -n 0

D3EAnalysis.py infile.out infile_ctrl.out outfile 1

I have tried it with t 0 and t 3.

It is a set of around 10 000 genes so I am running the job in parallel.
I do this by splitting the original table big_infile:

D3ESplitData.py big_infile Group1 Group2 ./ 300 1

This will create a few smaller tables which are prefixed Group1_Group2_.
For each of them I run the sequence described above.
Then I merge the resulting outfiles with:

D3EMergeOutput.py ./ final.out Group1 Group2

Thanks for making the code accessible! I found a potential bug in the '_readHeader function in D3EUtil.py. The number of columns are enumerated by the number of tabs in the first row but, typically, the row ends with the last label, not a tab, so the second group is always one short of the number of columns in the file unless you insert a tab at the end of the header file.

To be deleted