Open qiyunzhu opened 7 years ago
1) The CONDA_ENV_PATH issue I think has to do with the Conda version -- in the newer Conda it is stored as CONDA_PREFIX instead. I will make a note.
2) Thanks, forgot about BIOM -- will add that to the suggested install.
3) Will also add DIAMOND to the install
4) This last error I think has to do with there not being enough identifiable reads in the test data, and HUMANn2 not handling gracefully in these cases. The devs for HUMANn2 say that they are pushing a new version to PyPi soon that should fix this. Let me know if you still get this error with full-size data...
OK the first three should be fixed by commit https://github.com/tanaes/snakemake_shotqual/commit/b044c5578fe234086623f7f9b5c18a4b7520090d
@tanaes Thank you! Let me try your update.
Hello @tanaes , I tested your pipeline on some data. Here is my experience:
When executing raw_make_links_pe, it said:
Unable to set utime on symlink data/sample1/Run1/raw/sample1_R1.fq.gz. Your Python build does not support it.
The program needs an environment variable,
CONDA_ENV_PATH
. But it is not set yet. I had to manually set it:export CONDA_ENV_PATH=/home/me/path/to/conda/envs/kneaddata
In the pipeline HUMANN2 generates BIOM format output. This function requires BIOM, which is not installed by HUMANN2 by default. Therefore I had to install it manually:
pip install biom-format; conda install h5py
The latest HUMANN2 (0.9.5) needs DIAMOND 0.8.22, but it didn't automatically install it. I guess the reason may be that I already have DIAMOND 0.8.20 installed and added to the environment variable. The HUMANN2 installation program detected this and skipped DIAMOND installation. However, when using HUMANN2 it said that I should upgrade 0.8.20 to 0.8.22.
Even after these were down, HUMANN2 still couldn't run through. The first error is
biom.exception.TableException: Number of observation IDs differs from matrix size!
I have sent your the output files via email.