#
#	Species related functions (mostly related to SwissProt)
#
#					Gaston H. Gonnet (Dec 17, 2001)

Species_Entry := proc( specname:string )
global SearchOS_table, Species_table;

if not type(DB,database) then
     error('DB is not assigned to any database.');
fi;
if not type(Species_table,list(posint)) then
     t := CreateArray(1..DB[TotEntries],'');
     for i to DB[TotEntries] do
         t[i] := SearchTag('OS',DB[string,DB[entry,i]+1..DB[entry,i+1]]) od;
     SearchOS_table := sort(t);

     Species_table := [1];
     for i to length(SearchOS_table)-1 do
	 if SearchOS_table[i] <> SearchOS_table[i+1] then
	     Species_table := append(Species_table,i+1) fi od;
     Species_table := append(Species_table,i+1);
     fi;

found := [];
for i to length(Species_table)-1 do
     if SearchString( specname, SearchOS_table[Species_table[i]] ) >= 0 then
	 found := append(found,i) fi
	 od;
if found=[] then for tol to 3 do
     for i to length(Species_table)-1 do
	 if ApproxSearchString(specname,SearchOS_table[Species_table[i]],tol)
		>= 0 then found := append(found,i) fi
	 od;
     if length(found) > 0 then break fi;
     od fi;

r := [];
for i in found do
     for j from Species_table[i] to Species_table[i+1]-1 do
	 r := append(r,Entry(SearchOS_table[j]))
	 od;
     if printlevel > 1 then
	 printf( '%3d sequences of %s\n', Species_table[i+1]-Species_table[i],
	     SearchOS_table[Species_table[i]] ) fi
     od;

r
end:



#
#	Extract names of species from SwissProt
#
#	Optionally, select the species having a particular
#	taxonomic classification 
#					Gaston H. Gonnet (Dec 18th, 2001)
SP_Species := proc( taxon:string )
# Setup tables
if not type(DB,database) then
    error('a SwissProt database must be loaded in DB') fi;
Species_Entry('Abies veitchii');
if not type(Species_table,list(posint)) then
    error('could not create Species_table') fi;

Species := [];
for i to length(Species_table)-1 do
    t := SearchOS_table[Species_table[i]];
    if nargs=1 then
	if SearchString( taxon, SearchTag( 'OC', Entry(t) )) < 0 then next fi
	fi;
    j := SearchString( ' (', t );
    if j > 0 then t := t[1..j] fi;
    j := SearchString( ', and', t );
    if j > 0 then t := t[1..j] fi;
    while member( t[-1..-1], {' ', '.'}) do t := t[1..-2] od;
    Species := append(Species,t)
    od:
{op(Species)}
end:




SPCommonName := proc( entry )
if type(entry,{posint,structure}) or type(entry,string) and entry[1..3]='<E>'
     then en := Entry(entry);  t := SearchTag('OS',en);
elif type(entry,string) then
     eo := GetOffset(entry);
     if eo < 0 or eo >= DB[TotChars] then
	 eo := SearchString( entry, DB[string] );
	 if eo < 0 then error('invalid argument') fi;
	 fi;
     t := eo+DB[string];
     i := SearchString('</',t);
     if i >= 0 then t := t[1..i] fi;
     en := ''
else error('invalid argument, must be entry, entry number or description') fi;

if t <> '' then
     i := SearchString('(',t);
     if i < 0 then
	  while member(t[-1..-1],{' ','.'}) do t := t-1 od;
	  return( t ) fi;
     t := i+1+t;
     i := SearchString(')',t);
     if i < 0 then return(t) else return(t[1..i]) fi
     fi;

for z in ['AC','ID'] do
     t := SearchTag(z,en);
     if t <> '' then return(t) fi;
     od;

t
end: