You need to change how the matcher functionality is presented to the user, therefore you'll need two new ACD files and a copy of the matcher source code for each. Both new applications will effectively use the same source code with only trivial differences.

The steps to create these applications are as follows:

The original matcher.acd is shown below (help: attributes are not shown):

application: matcher 
[
  documentation: "Waterman-Eggert local alignment of two sequences"
  groups: "Alignment:Local"
]

section: input 
[
  information: "Input section"
  type: "page"
]

  sequence: asequence  
  [
    parameter: "Y"
    type: "any"
  ]

  sequence: bsequence  
  [
    parameter: "Y"
    type: "@($(acdprotein) ? stopprotein : nucleotide)"
  ]

  matrix: datafile  
  [
    additional: "Y"
    information: "Matrix file"
    protein: "$(acdprotein)"
  ]

endsection: input

section: additional 
[
  information: "Additional section"
  type: "page"
]

  integer: alternatives  
  [
    additional: "Y"
    information: "Number of alternative matches"
    default: "1"
    minimum: "1"
  ]

  integer: gapopen  
  [
    additional: "Y"
    information: "Gap penalty"
    default: "@($(acdprotein)? 14 : 16)"
    minimum: "0"
    valid: "Positive integer"
    expected: "14 for protein, 16 for nucleic"
  ]

  integer: gapextend  
  [
    additional: "Y"
    information: "Gap length penalty"
    default: "@($(acdprotein)? 4 : 4)"
    minimum: "0"
    valid: "Positive integer"
    expected: "4 for any sequence"
  ]

endsection: additional

section: output 
[
  information: "Output section"
  type: "page"
]

  align: outfile  
  [
    parameter: "Y"
    aformat: "markx0"
    minseqs: "2"
    maxseqs: "2"
  ]

endsection: output

The ACD file introduces several new concepts:

The changes necessary for matcherpro.acd are:

The parts of matcherpro.acd which have been modified and differ from matcher.acd are shown below:

application: matcherpro [
  documentation: "Waterman-Eggert local alignment of two sequences"
  groups: "Alignment:Local"
]

... lines omitted  

  sequence: asequence  [
    parameter: "Y"
    type: "protein"
  ]

  sequence: bsequence  [
    parameter: "Y"
    type: "stopprotein"
  ]

... lines omitted  


  matrix: datafile  [
    additional: "Y"
    information: "Matrix file"
    protein: "Y"
  ]

... lines omitted  

  integer: gapopen  [
    additional: "Y"
    information: "Gap penalty"
    default: "14"
    minimum: "0"
    valid: "Positive integer"
    expected: "14"
  ]

  integer: gapextend  [
    additional: "Y"
    information: "Gap length penalty"
    default: "4"
    minimum: "0"
    valid: "Positive integer"
    expected: "4 for any sequence"
  ]

... lines omitted  

The main() function for matcher is shown below. The application includes several functions and macros that are not shown:

#include "emboss.h"


/* @prog matcher **************************************************************
**
** Finds the best local alignments between two sequences
**
******************************************************************************/

int main(int argc, char **argv)
{
    AjPStr aa0str = 0;
    AjPStr aa1str = 0;
    const char *s1;
    const char *s2;
    ajint gdelval;
    ajint ggapval;
    ajuint i;
    ajint K;

    AjPAlign align = NULL;

    embInit("matcher", argc, argv);

    seq     = ajAcdGetSeq("asequence");
    ajSeqTrim(seq);
    seq2    = ajAcdGetSeq("bsequence");
    ajSeqTrim(seq2);
    matrix  = ajAcdGetMatrix("datafile");
    K       = ajAcdGetInt("alternatives");
    gdelval = ajAcdGetInt("gapopen");
    ggapval = ajAcdGetInt("gapextend");
    align   = ajAcdGetAlign("outfile");


    /*
      create sequence indices. i.e. A->0, B->1 ... Z->25 etc.
      This is done so that ajBasecodeToInt has only to be done once for
      each residue in the sequence
    */

    ajSeqFmtUpper(seq);
    ajSeqFmtUpper(seq2);

    s1 = ajStrGetPtr(ajSeqGetSeqS(seq));
    s2 = ajStrGetPtr(ajSeqGetSeqS(seq2));

    sub = ajMatrixGetMatrix(matrix);
    cvt = ajMatrixGetCvt(matrix);


    aa0str = ajStrNewRes(2+ajSeqGetLen(seq)); /* length + blank + trailing null */
    aa1str = ajStrNewRes(2+ajSeqGetLen(seq2));
    ajStrAppendK(&aa0str,' ');
    ajStrAppendK(&aa1str,' ');

    for(i=0;i<ajSeqGetLen(seq);i++)
        ajStrAppendK(&aa0str,(char)ajSeqcvtGetCodeK(cvt, *s1++));

    for(i=0;i<ajSeqGetLen(seq2);i++)
        ajStrAppendK(&aa1str,ajSeqcvtGetCodeK(cvt, *s2++));

    matcher_Sim(align, ajStrGetPtr(aa0str),ajStrGetPtr(aa1str),
                seq,seq2,
                K,(gdelval-ggapval),ggapval,
                ajSeqGetOffset(seq), ajSeqGetOffset(seq2), 2);

    ajStrDel(&aa0str);
    ajStrDel(&aa1str);

    ajSeqDel(&seq);
    ajSeqDel(&seq2);

    embExit();

    return 0;
}

matcher.c introduces many new concepts:

One small edit is needed in the src/matcherpro.c file:

The parts of matcherpro.c which have been modified and differ from matcher.c are shown below:

/* @prog matcher **************************************************************;

embInitP("matcherpro", argc, argv, "myemboss");

To compile and install the application from the myemboss directory, type:

Or, if you are using myemboss as part of a fully installed EMBOSS system then type:

matcherpro has just been added to your system, therefore you may need to make it visible in your path, for example, by running the UNIX command rehash, if you use a csh shell.

To check all is well, you can run:

and check that no errors or warnings are reported. As matcher.acd has already been validated it should be fine.

You should now have a new application matcherpro which should produce the same results as matcher with protein input, but will refuse to run with DNA sequences. This is shown below:

% matcherpro
Finds the best local alignments between two protein sequences
Input protein sequence: embl:L07770
Error: Sequence is not a protein

Error: Unable to read sequence 'embl:L07770'
Input protein sequence:
...

Here it is being used correctly:

% matcherpro
Finds the best local alignments between two protein sequences
Input protein sequence: swissprot:UBR5_RAT
Second protein sequence:
...

From the above it should be obvious what's required to implement matchernuc.acd and matchernuc.c, so generating two distinct applications, one for nucleotide and one for protein sequences.

If you are familiar with EMBOSS, you might notice that the output format of matcherpro is not the same as water, another popular alignment program. This could be confusing to users, so as a final modification you can make matcherpro consistent with water. To do this, set the aformat: attribute of the outfile data definition, which currently has the value "markx0" to the value ("srspair") defined in the water.acd file.

A similar effect could be achieved by running matcherpro with -aformat srspair on the command line. If for some reason the old format was required, matcherpro could be run with -aformat markx0 on the command line.