new de tag with samtools

[nikosdarzentas]

Hello,
and it's an understatement to say thank you for your work...

While testing the latest minimap2 2.15-r905 with fgbio, fgbio's SAM parsing complained that it was expecting a float from this new tag you added, with the offending example being:
de:f:-inf
Sorry in advance, I could only recover some of the log output from my terminal, hope it still helps:

Exception in thread "main" htsjdk.samtools.SAMFormatException: Error parsing text SAM fi[...]
        at htsjdk.samtools.SAMLineParser.reportErrorParsingLine(SAMLineParser.java:468)
        at htsjdk.samtools.SAMLineParser.parseTag(SAMLineParser.java:423)
        at htsjdk.samtools.SAMLineParser.parseLine(SAMLineParser.java:346)

[lh3]

Could you show me the offending line? Thanks.

[lh3]

PS: I haven't found a de:f:-inf in my local SAM/PAF files, so this is probably a rare bug. In this case, could you send me the query sequence? If you can't share the sequence, please send me via email. Thank you!

[nikosdarzentas]

Apologies, I had moved on by then and had to go back and recreate the analysis.
So, this is alignment with minimap2 -xsr of the sequence that is shown below (noisy, I know) to Ensembl's hg38 primary assembly genome, fed to fgbio, which produces the error - I hope that's adequate, but please get back to me if you need more.

[...]
[2019/01/11 21:41:35 | FgBioMain | Info] Executing SortBam from fgbio version 0.7.0 as nikos@core on JRE 1.8.0_121-b13 with snappy
[2019/01/11 21:41:36 | FgBioMain | Info] SortBam failed. Elapsed time: 0.14 minutes.
Exception in thread "main" htsjdk.samtools.SAMFormatException: Error parsing text SAM file. Tag of type f should have single-precision floating point value; File /dev/stdin; Line 87038
Line: NB501229:189:HCHKVAFXY:2:11105:26796:6432	163	18	68537384	60	150M	=	68537661	427	GTTTTTATACTGTATCATTTAGGGAATAATNATGNNNNNAAGNNNATNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNANNNNNNANNNNAGAACCCACAGGTAAAAATGACCAACTGTATTTGC	AAAAAEEEEEEEEEEEEEEEEEEEEEEEEE#EEE#####EEE###EE########################################################E######/####EAAAE6<AEAAAAEEEA<AA/<<AAE<AEEA<EE/	NM:i:75	ms:i:75	AS:i:225	nn:i:75	tp:A:P	cm:i:6	s1:i:178	s2:i:0	de:f:-inf	RX:Z:AATGGCAGTA
	at htsjdk.samtools.SAMLineParser.reportErrorParsingLine(SAMLineParser.java:468)
	at htsjdk.samtools.SAMLineParser.parseTag(SAMLineParser.java:423)
	at htsjdk.samtools.SAMLineParser.parseLine(SAMLineParser.java:346)
	at htsjdk.samtools.SAMTextReader$RecordIterator.parseLine(SAMTextReader.java:268)
	at htsjdk.samtools.SAMTextReader$RecordIterator.next(SAMTextReader.java:255)
	at htsjdk.samtools.SAMTextReader$RecordIterator.next(SAMTextReader.java:228)
	at htsjdk.samtools.SamReader$AssertingIterator.next(SamReader.java:576)
	at htsjdk.samtools.SamReader$AssertingIterator.next(SamReader.java:548)
[...]

[lh3]

Thanks for the example. I have fixed this bug. Nonetheless, fgbio is not doing the right thing, either. "-inf" is a valid floating point number.

[nikosdarzentas]

Great, many thanks! And should I tell them? It'll hurt their feelings... :-)

[jmarshall]

-inf is not a valid floating point number in SAM: someone named lh3 wrote in the specification that type f tag values must match the regular expression [-+]?[0-9]*\.?[0-9]+([eE][-+]?[0-9]+)?.

There is something to be said for relaxing this to allow for infinities and NaNs. If you would like to see this happen, please raise it as an issue in hts-specs.