bug: run_differential_usage.R - undefined columns selected with rowMeans

[SamBryce-Smith]

Log contains an error message like the following:

Performing differential usage analysis... Filtering single-isoform genes... Number of isoforms pre-filtering - 44132 Number of isoforms post-filtering - 38724 Filtering isoforms for minimum mean count in any condition of at least - 10 Error in h(simpleError(msg, call)) : error in evaluating the argument 'x' in selecting a method for function 'rowMeans': undefined columns selected Calls: dtu_wrapper ... sapply -> lapply -> FUN -> rowMeans -> [ -> [.data.frame Execution halted

Here, columns are sample_names in the counts matrix. This seems to happen when sample_name values contain hyphens ('-'), otherwise the pipeline sails through.

To-do:

• Add an explicit check that sample_name values in sample table cannot contain '-' before pipeline runs
• Document this requirement in the config and README.

[SamBryce-Smith]

Fixed with 0c52d38a0a91d2ddb0acbb0692b2f2b1fcc86d4f & b6281eab54bd8746df35ddf81dc7e13c5fb7561b