Calculate phiX error rates from Undetermined file

[donkirkby]

We currently censor some tiles and cycles based on poor error rates from the interop files. There is currently no way to upload that interop data to BaseSpace from an old run, it can only be uploaded during a run.

Casey at Illumina suggested we could recalculate the phiX error rate data by aligning the reads to a phiX reference sequence. The phiX reads aren't labelled, so they are probably all in the Undetermined.fastq file.

• See if an Undetermined.fastq file will upload to BaseSpace.
• Run an Undetermined.fastq file through the current MiCall pipeline to see what it finds. This is just for curiosity.
• Add a phiX reference to the pipeline, with a coordinate reference. See what it maps from some samples and an Undetermined.fastq file.
• Recalculate the error rates, and compare them to the values in the interop files.
• Rewrite the pipeline to use Undetermined.fastq instead of the interop files. Consider splitting into two separate pipelines to avoid running the quality step more than once.

[donkirkby]

The undetermined FASTQ file has a sample number 0 in each read's header. That's for FASTQ files from the MiSeq. Other instruments put the sample's nucleotide barcode there, so I'm not sure what the undetermined file would have.

[donkirkby]

We could make sure the Undetermined file is from the same run as the sample files by checking the flow cell id at the end of each read header.