Restriction in read name length

Question

[grant.hovhannisyan]

Hi all,

Does anybody know if there is a restriction for a read name length (say up to 10k characters)? Have you ever experienced that any software (like samtools, bwa, STAR, etc) complains about long read names?

Thanks

[ATpoint]

Please define long. 100 characters, 10000 characters?

[grant.hovhannisyan]

See the edited version.

[Pierre Lindenbaum]

what's your motivation for doing this ?

[grant.hovhannisyan]

I am writing a pipeline, where there is a read simulation step, and I want to include some information in the read name, like chr name (which can be long in some cases), coordinates, strand etc.

Answer 1

[Pierre Lindenbaum]

in htslib / sam.h https://github.com/samtools/htslib/blob/develop/htslib/sam.h

the read name is coded by

/* @field  l_qname length of the query name */
 uint8_t l_qname;

which is an unsigned byte. So the read name shouldn't be longer than 2^8=256.

it's even shorter in the read-sam method:

https://github.com/samtools/htslib/blob/f2150106bafda80ccd97970637091bb2799bc426/sam.c#L1305

 _parse_err(p - q > 252, "query name too long");

[grant.hovhannisyan]

Thanks, very helpful.