search

FoxoTech / methylprep

Python-based preprocessing software for Illumina methylation arrays
MIT License
34 stars 14 forks source link

missing data in manifest files for non-CpG probes; FIX requires users to delete their local manifest files and methylprep will redownload them. #41

Closed marcmaxson closed 4 years ago

marcmaxson commented 4 years ago

Nichole Rigby - 9:17 AM DEC 12 Not sure if this is a bug or not, but a single control probe shows up in the methylprep version of the manifest. Its one of the -99 ones. I'm guessing it should be in the control dataframes, and not the manifest?

Another bug, (and this effects our new QC functions), is that some of the controls are missing from the control dataframes. For example, the raw manifest (outside of methylprep) contains these STAINING controls.

New bug related to the control probes: I also found some inconsistency in the names of the BISULFITE CONVERSION probes. Some platforms have names like BS Conversion I-C1 (EPIC+) and others have similar names with the - missing (450K) for only SOME of the BISULFITE CONVERSION probes. In some cases, the original manifest has names with the - and the manifest in methylprep is missing the - from the name.

9:09 AM I think the solution might be to through all the manifest files and methylprep manifests and make sure the - is in the name for each.

Nichole: I checked them all, only changed and re-uploaded the ones that were affected also had to delete the .methylprep_manifest_files directory. My code now works.

For users to get the update, they'll need to delete the manifest files on their machine, and it will re-download them.

marcmaxson commented 4 years ago

Related bug: processing samples with the --save_control flag won't recognize EPIC+ array control probes. The values are all NAN. But SNP probe data is found in the control_probes.pkl dataframes.