[EMBOSS] emma

[Peter Rice]

Staffa, Nick (NIH/NIEHS) wrote:
> ******** CLUSTAL 2.0.7 Multiple Sequence Alignments  ********
> 
> Command line emma in conjunction with this program generates this error:
> 
> CLUSTAL 2.0.7 Multiple Sequence Alignments Error: parameter
> required for /dnamatrix Error: Failed to open filename '00003189B'

There is a bug in emma, not noticed before because Clustalw 1.x 
tolerated the command line.

In emboss/emma.c the code for option dnamatrix should be:

     dna_matrix = ajAcdGetListSingle( "dnamatrix");
     m2c = ajStrGetCharFirst(dna_matrix);

     if(m2c=='i')
         ajStrAssignC(&m2str,"iub");
     else if(m2c=='c')
         ajStrAssignC(&m2str,"clustalw");
     else if(m2c=='o')
         ajStrAssignC(&m2str,"own");

(the characters tested for m2c were from another command line option so 
the dnamatrix option was not being passed to clustalw).

Clustalw 1.8.x ignored the empty /dnamatrix string on the commmand line.

> Y'd think the defaults would be sufficient.

They are now :-)

> Y'd think that Jemboss would do a better job of reporting problems.

It is always tricky when the error is reported by another application 
invoked through EMBOSS.

My favourite emma error was a Taverna workflow that failed when 
executing an emma webservice. Clustalw crashed. The cause turned out to 
be a user-generated FASTA file with bad end of line characters in the 
sequence so the sequences became part of the first ID and emma launched 
clustalw with a set of zero length sequences. We "fixed" it by requiring 
all EMBOSS sequences to have at least one valid character so we can now 
issue an EMBOSS error.

regards,

Peter