Petr Danecek

Results 426 comments of Petr Danecek

The test files are not valid VCF format. Please provide a tarball for testing, that can be created with something like ``` tar -czf test.tar.gz input*.vcf ```

I believe it is supported, only needs to be named as `something.tar.gz`, as I just tested. I edited the command above to reflect that

> I am not sure I agree with the 'without public discussion' part; we had two release candidates of 4.5 out for several months and requested comments; notes went out...

> Quite the opposite. END was deprecated in favour of FORMAT LEN as END is incompatible with per-sample gVCF (``) reference blocks. This change was included as part of the...

> There were attempts to raise awareness of VCF 4.5 release candidates on github, in a GA4GH email circular IIRC, and also via twitter (eg https://twitter.com/camerongenomics/status/1815619781199892971, including INFO/END, reposted numerous...

The very wording "is present for the purposes of backwards compatibility with older software" is what I am having a problem with and what this issue is about. My question...

That is correct in that the indexes we have so far are purely site-oriented, not sample-oriented. If the indexing is extended in the future to support individual samples, then parsing...

@jkbonfield The confusing language of END being "deprecated" is influencing programmers' decisions, please see for example here: https://github.com/freeseek/score/issues/35#issuecomment-3049531958. Here, the program deliberately ignores INFO/END. Can this be fixed, please? It...

> I'm not sure why you're tagging me as I'm not the one who deprecated it and nor am I the specification maintainer. @jkbonfield Sorry, you were most active in...

@d-cameron thank you for this. From a practical point of view, it would be better to phrase it so as to make INFO/END a "derived" field, rather than "deprecated", to...