bool open_weights(
const char *weight_file,
s_motif *motifs[],
int num_motifs )
{
// open the file
ifstream ifs;
_CALL(
_open_input( ifs, weight_file, ".txt" )
);
// find the EOL character
char this_EOL = _find_EOL( ifs );
if( !this_EOL )
// no EOL character was identified
{
cout << "ERROR: The file has no EOL character." << endl;
return false;
}
// read the file
cout << "Reading the weights..." << endl;
_CALL(
read_weights( ifs, motifs, num_motifs, this_EOL )
);
ifs.close();
return true;
}
MYI_STRUCT *_fancy_open_input(NFN_STRUCT *ESI, FILE_LISTS *ECX)
{
// ESI is NFN_STRUCT containing raw filename
// ECX is list of paths
if (!_open_input(ESI, &OBJ_DEVICE))
return OBJ_DEVICE;
// Well, it would have been nice
if (ESI->NFN_FLAGS & NFN_PATH_SPECIFIED)
goto FANCY_FAIL;
FILE_LIST_STRUCT *EDI = ECX->FILE_FIRST_GINDEX;
goto TEST_PATH;
PATH_LOOP:
if (EDI->FILE_LIST_NFN.NFN_FLAGS & NFN_PATH_SPECIFIED) // skip 'NUL'
{
// Need to move just path
if (!_open_input(_move_ecxpath_eax(ESI, &EDI->FILE_LIST_NFN), &OBJ_DEVICE))
{
FILE_LIST_STRUCT *EDX = CURN_FILE_LIST_GINDEX;
if (EDX)
EDX->FILE_LIST_PATH_GINDEX = EDI;
return OBJ_DEVICE;
}
}
// Try next path
TEST_PATH:
EDI = EDI->FILE_LIST_NEXT_GINDEX;
if (EDI)
goto PATH_LOOP;
FANCY_FAIL:
_delete_phantom_path(ESI);
return 0;
}
bool open_FASTA(
const char *file,
s_seq *seqs[],
int *num_seqs )
// opens a file and reads its content
// returns true if successful, false otherwise
//
{
// open the file
ifstream ifs;
_CALL(
_open_input( ifs, file, ".fasta" )
);
// find the EOL character
char this_EOL = _find_EOL( ifs );
if( !this_EOL )
// no EOL character was identified
{
cout << "ERROR: The file has no EOL character." << endl;
return false;
}
// read the file
cout << "Reading the sequences..." << endl;
_CALL(
read_FASTA( ifs, seqs, num_seqs, this_EOL )
);
// count the number of positive instances
int num_pos = 0;
int i;
for( i = 0; i < *num_seqs; i ++ )
num_pos += seqs[ i ] ->positive;
cout << *num_seqs << " sequences were read, including " << num_pos << " positive and "
<< *num_seqs - num_pos << " negatives." << endl << endl;
if( num_pos < 10 || *num_seqs - num_pos < 10 )
{
cout << "ERROR: At least 10 positive and 10 negative sequences are required." << endl;
return false;
}
// close the input file stream
ifs.close();
return true;
}
bool open_rndforest(
const char *file,
s_motif *motifs[],
int *num_motifs,
char const *filter )
// opens a file and reads its content
// returns true if successful, false otherwise
{
// open the file
ifstream ifs;
_CALL(
_open_input( ifs, file, ".txt" )
);
// find the EOL character
char this_EOL = _find_EOL( ifs );
if( !this_EOL )
// no EOL character was identified
{
cout << "ERROR: The file has no EOL character." << endl;
return false;
}
// read the file
cout << "Reading RF results..." << endl;
_CALL(
read_rndforest( ifs, motifs, num_motifs, filter, this_EOL )
);
if( *num_motifs )
cout << *num_motifs << " motifs were read." << endl << endl;
else
{
cout << "ERROR: No motifs were read. The input protein possibly does not have canonical C2H2-ZFs." << endl;
return false;
}
// close the input file stream
ifs.close();
return true;
}
