Sample sheet parser fails when enum contains an underscore

[CBeelen]

Our sample sheet parser uses underscores to separate enum from project code in the sample name, and when unpacking the additional information in the description column. When there is an underscore present in the enum itself, the parser will use the wrong project code, and it will fail at a later point when unpacking the description, because it ends up with too many values when splitting the sample name at underscores.

Possible solutions:

• Make the sample sheet parser more robust to this - that seems a little difficult and clunky
• Disallow users from typing underscores as part of an enum in QAI (currently it looks like the enum can be anything, see /qcs_miseq_tag/save in the qcs_miseq_tag_controller

David's suggestion of using a regex to parse the sample sheet won't really work unfortunately, the parsing code is pretty complex and covers a lot of different layouts and scenarios.

Unfortunately, the sample sheet override option won't help us here (I think), because it will only do the override once the sample sheet has been read in successfully.

[donkirkby]

For the short-term goal of processing that run, I think you might get away with just editing the sample sheet to remove the underscore and then redoing the demultiplex step. When MiCall Watcher uploads the results to QAI, it links them based on the tags, not the sample name. As long as the tags match, I think we can get away with renaming samples.

[CBeelen]

QAI would be happy with that, yes - but miseq_gen_results.rb checks the enums against the entries in the database in specimen.lab_miseq_tagged_layout_samps, so that would be unhappy when we change the sample name. In this particular case it might be fine, because the project code is Unknown anyways, I think.