HML Spec: What should raw-reads look like?

[dvaliga]

What data formats/options do DaSH attendees see sending as 'raw reads' to NMDP? FASTA? FASTQ? Done as an HML message payload or external link? If URI, where will the data live? For how long?

[mmaiers-nmdp]

According to the MIRING the raw reads are not part of the message so it will not appear in HML.

[gturenchalk]

According to the MIRING Documentation the primary reads are covered as Category 9 data of an HGS/HTS-specific "Accessory Data" addendum to the main MIRING message in the genotyping report.
My interpretation is that a URI will be provided to raw data in SFF or FASTQ format that has been uploaded to the SRA. Others, please correct me if I am misinterpreting. The excerpt on Category 9 data is provided below:

Category 9:
Primary Data: unmapped reads with quality scores must be made
available as the primary NGS data, permitting re-analysis of the genotype
result by different NGS analytic software. This primary data must be
limited to full-length reads that include syntactically valid adapter and
indexing/barcoding sequences. However, adapter sequences need not be
included in the primary data.
Due to their potential large size, it may not be possible to make the
primary data available as part of a genotyping report; however, these data
must be made available through other electronic means (e.g., deposition
in the NCBI’s Sequence Read Archive), and instructions for obtaining
them must be included in the genotyping report.
We recommend using either the Sanger FASTQ or Standard Flowgram
Format (SFF, used by the Roche 454 platform) or a comparable format to
report unmapped sequence reads for NGS HLA or KIR primary data. SFF
files can be converted to FASTQ format using available software.