# the function WriteFastaWithRefs is removed and merged with WriteFasta.  AR (April 2007)

WriteExpStrucFile := proc(filename:string, expstruc:ExpStruc)

if not assigned(ExpStruc) then ReadLibrary ('ExpStruc') fi:

  OpenWriting(filename):

   printf('%s\n','if not assigned(ExpStruc) then ReadLibrary (''ExpStruc'') fi:'):
   printf('%s\n', 'es := ExpStruc ():'):
   printf('\n'):

   if es['Seq'] <> 0  then
     printf('%s','es[''Seq''] := '):
     printf('%s:', es['Seq']):
     printf('\n\n'):
   fi:

   if es['SecStr'] <> 0  then
     printf('%s','es[''SecStr''] := '):
     printf('%a:', es['SecStr']):
     printf('\n\n'):
   fi:

   if es['Target'] <> 0  then
     printf('%s','es[''Target''] := '):
     printf('%d', es['Target']):
     printf('%s\n\n',':'):
   fi:

   if es['SI'] <> 0  then
     printf('%s','es[''SI''] := '):
     printf('%s', es['SI']):
     printf('%s\n\n',':'):
   fi:

   if es['PDBnum'] <> 0  then
     printf('%s','es[''PDBnum''] := '):
     printf('%a', es['PDBnum']):
     printf('%s\n\n',':'):
   fi:

   if es['MV'] <> 0  then
     printf('%s','es[''MV''] := '):
     printf('%a', es['MV']):
     printf('%s\n\n',':'):
   fi:

   if es['Topo'] <> 0  then
     printf('%s','es[''Topo''] := '):
     printf('%a', es['Topo']):
     printf('%s\n\n',':'):
   fi:

   if es['SurfAcc'] <> 0  then
     printf('%s','es[''SurfAcc''] := '):
     printf('%a', es['SurfAcc']):
     printf('%s\n\n',':'):
   fi:

   if es['Naomi'] <> 0  then
     printf('%s','es[''Naomi''] := '):
     printf('%a', es['Naomi']):
     printf('%s\n\n',':'):
   fi:

 OpenWriting(terminal):

end:

WriteParseFile := proc (filename:string, hp:array(array(array(integer))), vp:array(array(integer)))

OpenWriting(filename):

   vlen_seq := length(vp[1]):

    s := CreateString (vlen_seq, ' '):

    for i to length (vp) do
      for pos to vlen_seq do
        for j to vp[i,pos] do
          s[pos+j-1] := '*'
        od
      od
    od:

    printf ('%s\n','vpstring := '):
    printf ('%s\n',s):
    printf ('%s\n',':'):

    num_seq := length(hp[1]):
    len_seq := length(hp[1,1]):

    hparse := CreateArray (1..num_seq,1..len_seq):

    for i to length (hp) do 			# loop over hp rules
      for seq to num_seq do 			# loop over sequences
        for pos to len_seq do 			#loop over positions
          if hp[i,seq,pos] <> 0 then
            for j to hp[i,seq,pos] do
             hparse[seq,pos + j -1] := 1:
            od:
          fi
        od
      od
    od:

    printf ('%s\n','hparse := '):
    printf ('%a\n',hparse):
    printf ('%s\n',':'):

    printf ('%s\n','hp := '):
    printf ('%a\n',hp):
    printf ('%s\n',':'):

    printf ('%s\n','vp := '):
    printf ('%a\n',vp):
    printf ('%s\n',':'):

OpenWriting(terminal):

end:

ReadClustal := proc(fname:string)

  ReadLibrary ('MBA_Toolkit');

  # -- processing optional arguments --

  compressed := false;

  for i from 2 to nargs do
    if type (args[i], string = anything) then
      if op(1, args[i]) = 'Compressed' and type (op (2, args[2]), boolean) then
        compressed := op (2, args[i])
      else
        print ('proc(' . procname . ') WARNING -- uncaught argument:', args[i])
      fi
    else
      print ('proc(' . procname . ') WARNING -- uncaught argument:', args[i])
    fi
  od;

  # -- initializations --

  if compressed then
    cmd := 'zcat'
  else
    cmd := 'cat'
  fi;

  OpenPipe (cmd . ' ' . fname);

#main loop

  body := false;
  first := true;
  second := false;
  refs := [];
  seqs := [];

  do
    t := ReadRawLine ();

    if t = EOF then
      break
    fi;

  if length(t) <> 0 then

  if t[1] = ' ' then 
   body := false ;
   first := false;
   second := true;
   count := 1;
  fi;
   
  if t[1] <> ' ' and SearchString('CLUSTAL',t) = -1 then
   body := true;
  fi;

  if body then

    end_of_ref := SearchString(' ',t);
    beg_of_ref := 1;
   
    for k from end_of_ref + 1 to length(t) do
     if t[k] <> ' ' then 
      break;
     fi;
    od;

    if first then
      refs := append(refs,t[beg_of_ref..end_of_ref]);
      seqs := append(seqs,t[k..length(t)]);
      ne := length(refs);
    fi;
  
    if second then
      seqs[count] := seqs[count] . t[k ..length(t)];
      count := count + 1;
    fi;

  fi;
 
  fi; # if length(t) <> 0 

  od;

  for i to length (seqs) do
    for j to length (seqs[i]) do
     if seqs[i,j] = '-' then
      seqs[i,j] := '_';
     fi;
    od;
  od;

  seq2 := GetRidOfGapsFromArray(seqs);

refs,seqs,seq2;

end:


ReadPretty := proc(fname:string)

  ReadLibrary ('MBA_Toolkit');

  # -- processing optional arguments --

  compressed := false;

  for i from 2 to nargs do
    if type (args[i], string = anything) then
      if op(1, args[i]) = 'Compressed' and type (op (2, args[2]), boolean) then
        compressed := op (2, args[i])
      else
        print ('proc(' . procname . ') WARNING -- uncaught argument:', args[i])
      fi
    else
      print ('proc(' . procname . ') WARNING -- uncaught argument:', args[i])
    fi
  od;

  # -- initializations --

  if compressed then
    cmd := 'zcat'
  else
    cmd := 'cat'
  fi;

  OpenPipe (cmd . ' ' . fname);

#main loop

  body := false;
  first := true;
  second := false;
  refs := [];
  seqs := [];

  do
    t := ReadLine ();

    if t = EOF then
      break
    fi;
   
  if SearchString('{',t) > 0 and SearchString('}',t) > 0 and SearchString('Pretty',t) = -1 then
   body := true;
  fi;

  if SearchString('Consensus',t) >0 then 
   body := false ;
   first := false;
   second := true;
   count := 1;
  fi;
   
  if body then

    end_of_ref := SearchString('}',t) ;
    beg_of_ref := SearchString('{',t) + 2;

    if first then
      refs := append(refs,t[beg_of_ref..end_of_ref]);
      seqs := append(seqs,t[end_of_ref + 3..length(t)]);
      ne := length(refs);
    fi;
  
    if second then
      seqs[count] := seqs[count] . t[end_of_ref + 4..length(t)];
      count := count + 1;
    fi;

  fi;

  od;
 
  seq2 := CleanSeq(seqs);

refs,seq2;

end:


GetRidOfGapsFromArray := proc(MA: array(string));
 n := length(MA); 
 Seq := CreateArray(1..n);
 for i to n do
   p := 0; 
   l := length(MA[i]);
   temp := '';
   for j to l do
     a := MA[i,j];
    if a < 'Z' and a >= 'A' then
     p := p + 1; 
     temp := temp . MA[i,j];
     fi;
   od;
   Seq[i]:=copy(temp[1..p]);
 od;
 Seq;
end:


CleanSeq := proc(MA: array(string));
 n := length(MA); Seq := CreateArray(1..n);
 for i to n do
   p := 0; l := length(MA[i]);
   temp := CreateString(l);
   for j to l do
     a := MA[i,j];
     if a <> ' ' then p := p + 1; temp[p] := MA[i,j]
#     else MA[i,j] := '_'; 
     fi;
   od;
   Seq[i]:=copy(temp[1..p]);
 od;
 Seq;
end:

WriteNexus := proc(seqs:array(string),refs:array(string),outfile:string)
  
  OpenWriting(outfile);

  ne := length(seqs);
  nr := length(seqs[1]);
  
  printf('#NEXUS\n');
  printf('BEGIN DATA;\n');
  printf('DIMENSIONS NTAX=%d NCHAR=%d;\n',ne,nr);
  printf('FORMAT MISSING=? GAP=_ DATATYPE=PROTEIN;\n');
  printf('MATRIX\n');
  
  for i to ne do
    printf('%s\n',refs[i]);
    printf('%s\n',seqs[i]);
  od;

  printf(';\n');
  printf('END;\n');

  OpenWriting(terminal);
  
end:

TranslatePretty := proc(infile:string,outfile:string)

  #ReadLibrary('Cannaro/Tools/mike.tools'):
  #TranslatePretty('/usr/people/cannaro/mike/MYG/PRETTY','/usr/people/cannaro/junk90');

  a := ReadPretty(infile);
  refs := a[1];
  seqs := a[2];

  WriteNexus(seqs,refs,outfile);

end:

CleanSeq := proc(MA: array(string));
 n := length(MA); Seq := CreateArray(1..n);
 for i to n do
   p := 0; l := length(MA[i]);
   temp := CreateString(l);
   for j to l do
     a := MA[i,j];
     if a <> ' ' then p := p + 1; temp[p] := MA[i,j]
#     else MA[i,j] := '_'; 
     fi;
   od;
   Seq[i]:=copy(temp[1..p]);
 od;
 Seq;
end:


WriteNexus := proc(seqs:array(string),refs:array(string),outfile:string)
  
  description 'Write nexus format. Input seqs:array,refs:array,filename:string';
  OpenWriting(outfile);

  ne := length(seqs);
  nr := length(seqs[1]);
  
  printf('#NEXUS\n');
  printf('BEGIN DATA;\n');
  printf('DIMENSIONS NTAX=%d NCHAR=%d;\n',ne,nr);
  printf('FORMAT MISSING=? GAP=. DATATYPE=PROTEIN;\n');
  printf('MATRIX\n');
  
  for i to ne do
    printf('%s\n',refs[i]);
    printf('%s\n',seqs[i]);
  od;

  printf(';\n');
  printf('END;\n');

  OpenWriting(terminal);
  
end:

TranslatePretty := proc(infile:string,outfile:string)

  #ReadLibrary('Cannaro/Tools/mike.tools'):
  #TranslatePretty('/usr/people/cannaro/mike/MYG/PRETTY','/usr/people/cannaro/junk90');

  a := ReadPretty(infile);
  refs := a[1];
  seqs := a[2];

  WriteNexus(seqs,refs,outfile);

end: