From: Alec W. <al...@br...> - 2009-02-23 14:53:55
|
Hi Tim, A couple of points: I'm assuming that you meant to write @RG in the example below rather than @RD. I.e. you just want to add some additional fields to the existing @RG record rather than creating a new header record type, right? The existing Java APIs support something similar, but not exactly matching what you outline below. The difference is that the current code does not support typed tags in the header lines. After splitting a header line on "\t", tokens after the @ token are split into two pieces on ':'. For some known tag types, the value is converted into something other than String, but anything unrecognized is stored as a String. When writing these values, the text format is produced simply by calling toString(). So, e.g. byte[] is not supported. I don't have a strong preference one way or the other, but we should nail down which way we want to go, and I'll fix the Java API if necessary. -Alec Tim Fennell wrote: > Hi All, > > In the spec we allow for user defined tags in the alignment/read > records. What do you think about also allowing optional fields in the > header records? For example I'd like to be able to add a couple more > tags into the read group header that are somewhat specific to our > pipeline (essentially embedding a couple of bits of data that will > make submissions easier). So I'd like to be able to write a header > record like: > > @RD ID:1 SM:foo PJ:Z:C123 WR:i:1111 > > Do you see any downside to allowing optional fields in the header > records? If not can we add a sentence to the specification that > states that this is allowed (and that parsers should deal with it)? > > Thanks, > > -t |