[emboss-dev] Specifying sequence lists for seqret

[Rimvydas Naktinis]

On 07/29/2010 03:28 PM, Peter Rice wrote:
> On 29/07/10 12:47, Rimvydas Naktinis wrote:
>> Hi,
>>
>> I'm developing EMBOSS integration into Chipster project
>> (chipster.csc.fi). I was wondering if there's a way to specify sequence
>> list (for example, when calling seqret) without creating any extra files?
>>
>> I know there's a way to do it like this:
>>>   seqret @sequencelist
>> or
>>>   seqret list:sequencelist
>>
>> But what I would need would look something like:
>>>   seqret "swiss:CASA1_RABIT,swiss:CASA1_HUMAN"
>>
>> I've looked into USA format specification and it seems that there is
>> actually no way to do it without creating some temporary file. Or am I
>> missing something?
>
> We can add that as an option ... but there is a very real danger that
> the command line will be too long.
>
> How many sequences will be on the command line (normal use, and worst case)?
>
> regards,
>
> Peter Rice

In current use case user enters the sequence names manually, so list 
should not be very long.

However, we should also think about a general case. As far as I know, 
starting from Linux kernel version 2.6.23 argv size is limited by 1/4th 
of memory stack size 
(http://www.kernel.org/doc/man-pages/online/pages/man2/execve.2.html), 
so argument length is basically limited only by available physical 
memory. Situation might be different with other operating systems. I 
guess in Windows the limit is 32Kb 
(http://msdn.microsoft.com/en-us/library/ms682425%28VS.85%29.aspx).

I guess this could be left for programmers to deal with and users, who 
use the command line themselves, would probably use the @seqlist option 
for long lists anyway.

And thanks for quick response!

Regards,
Rimvydas Naktinis
CSC — IT Center for Science Ltd