Merging samples causes sample attributes to be overwritten

[donaldcampbelljr]

Per some initial investigation, I think this is a looper bug not a peppy bug.

Given a sample table where the names are not unique:

sample_name,score,universe
sample1,score, cc
sample1,score, ccf
sample1,no_score, cc
sample1,no_score, ccf

I noticed that, when running looper, the attributes passed to my piface will be overwritten during the sample merge.

Found 1 samples with non-unique names: {'sample1'}. Attempting to auto-merge.

command template example

python3 pipeline/pipeline_parallel.py {looper.output_dir} {sample.sample_name} {sample.score} {sample.universe}

so sample.universe becomes the same value as whatever is sample.score

I thought maybe I needed to use subsample table for this particular set up but looking at and then utilizing a subsample table from the docs did not solve this.

I had to give each sample name a unique identifier else my attributes were overwritten.

[nleroy917]

Speculating and a hypothesis... the pepkit specification indicates that samples with the same sample_name will merge attributes into a list... and therefore maybe there is something in looper that sees that the attribute value is a list and then gets the first one so it appears its overwriting but its actually just a bug (like you suggested)