cytodata / cytodata-hackathon-2018 Goto Github PK

There are new folders called profiles_cp and they were presumably previously names profiles. Its possible this was done when we were reprocessing the data.

@jccaicedo Any clue what might have happened here?

Only Bioactives-BBBC022-Gustafsdottir has both profiles_cp and profiles. The others have only profiles_cp.

• check if all mean profiles are available
• run QC
  • for CP features: run the standard pipeline including feature selection and replicate correlation
  • for DL features: 1. run sanity check (make sure that all features have non zero values) and run similar tests feature selection and calculate replicate correlation

I'm trying to copy the BBBC022 profiles from cytodata s3 bucket to set up a benchmark pipeline for our cell painting method development. I can list the files in the bucket, but cannot copy or sync anything. Could you please change the permissions accordingly?
Thanks.

Treatments in BBBC037 are in the column Metadata_pert_name instead of Metadata_gene_name.

Aggregation code should be changed in the Python and R code examples.

The data sets are hosted and publicly available at

s3://cytodata/

All images, single cell features as sqlite files and aggregated features on replicate level can be found under

s3://cytodata/datasets/

A detailed description of all data is missing.

There are multiple versions of single cell features in cell painting gallery. Could you describe which processing tasks were done on them? I figured out the _augmented, which is features + metadata. Where could I find information about processing steps for the other files? Are they the same as in the source papers?

For example: _normalized - what kind of normalization was used?
_feature_select - what methodology was used?

Basically, any information that I could use to reproduce the processing steps used on the datasets.