[EMBOSS] Parameter discrepency

[pmr at ebi.ac.uk]

Hi Ryan,

> I tried -aformat and that seems to work, so does -aformat3.  Is one of
> them not supposed to be there or just a carryover from an older version
> of emboss?  The docs talk about one version but use the other in the
> example...

For all current programs, -aformat is fine.

To explain what is going on, consider the -osformat (output sequence
format) qualifier for a program that writes two sequences.

The more complex inputs and outputs have "associated qualifiers" in EMBOSS
- sequences for input, sequences, alignments, reports for output. We
usually make these parameters, as it is useful in scripts to simply put
the sequence "USA" or output filename on the commandline.

Simply using -osformat could apply to one or both sequences. There are
ways to specify one or both.

-osformat3 would apply to the third parameter (maybe one input sequence
and two output sequences - we try to insist on inputs first and outputs
last in the parameter order). The number is not recommended - as it would
change if we add or remove other qualifiers. However, the output of
"needle -help -verbose" is too long-winded if we use the full name (see
below) so we shorten it to -aformat3, -sformat1, -sformat2 for your needle
example)

Global settings:
-osformat at the start of the command line (strictly, before the first
parameter) applies to all.

Positional qualilfiers (an idea borrows from the VMS operating system):
-osformat after a parameter/qualifier that it is associated with applies
specifically to that parameter/qualifier

Full names:
-osformat_outseq refers to the -outseq parameter (all parameters also have
qualifier names). For needle, this would be aformat_outfile and is safer
than -aformat3

Normal users will use -aformat.

Wrappers and scripts should use -aformat_outfile (but can safely use
-aformat before specifying the output file)

Hope that helps to make things clearer.

Peter