static void CleanUpXOS (XOSPtr xosp)
{
SeqLocPtr slp, slpn;
SeqIdPtr id;
if (xosp != NULL)
{
xosp->gsp = NULL; /* is static */
xosp->filename = (CharPtr) MemFree (xosp->filename);
xosp->sep = SeqEntryFree (xosp->sep);
xosp->bsp = NULL; /* should be in seqentry */
xosp->gcd = NULL;
xosp->gcdi = NULL;
slp = xosp->slpa;
while (slp != NULL)
{
slpn = slp->next;
id = SeqLocId (slp);
if (id != NULL)
id->next = SeqIdSetFree (id->next);
SeqLocFree (slp);
slp = slpn;
}
xosp->slpa = slp;
slp = xosp->slpb;
while (slp != NULL)
{
slpn = slp->next;
id = SeqLocId (slp);
if (id != NULL)
id->next = SeqIdSetFree (id->next);
SeqLocFree (slp);
slp = slpn;
}
xosp->slpb = slp;
slp = xosp->slps;
while (slp != NULL)
{
slpn = slp->next;
id = SeqLocId (slp);
if (id != NULL)
id->next = SeqIdSetFree (id->next);
SeqLocFree (slp);
slp = slpn;
}
xosp->slps = slp;
slp = xosp->slpk;
while (slp != NULL)
{
slpn = slp->next;
id = SeqLocId (slp);
if (id != NULL)
id->next = SeqIdSetFree (id->next);
SeqLocFree (slp);
slp = slpn;
}
xosp->slpk = slp;
}
return;
}
extern void UniqueOrfs (SeqLocPtr PNTR pslpFound)
{
SeqLocPtr slpFound, slpNew, slp;
SeqIdPtr id;
slpFound = *pslpFound;
slpNew = NULL;
while (slpFound != NULL)
{
if (!SeqLocMatch (slpNew, slpFound))
SeqLocLink (&slpNew, SeqLocDup (slpFound));
slpFound = slpFound->next;
}
slpFound = *pslpFound;
while (slpFound != NULL)
{
slp = slpFound->next;
id = SeqLocId (slpFound);
if (id != NULL)
id->next = SeqIdSetFree (id->next);
SeqLocFree (slpFound);
slpFound = slp;
}
*pslpFound = slpNew;
return;
}
extern SeqLocPtr SeqLocDup (SeqLocPtr slpold)
{
SeqLocPtr slpnew, slpn, slp;
SeqIdPtr id;
if (slpold == NULL)
return NULL;
slpnew = (SeqLocPtr) AsnIoMemCopy ((Pointer) slpold,
(AsnReadFunc) SeqLocAsnRead,
(AsnWriteFunc) SeqLocAsnWrite);
slp = slpnew->next;
while (slp != NULL)
{
slpn = slp->next;
id = SeqLocId (slp);
if (id != NULL)
id->next = SeqIdSetFree (id->next);
SeqLocFree (slp);
slp = slpn;
}
slpnew->next = NULL;
return slpnew;
}
SBlastMessage* SBlastMessageFree(SBlastMessage* message)
{
SBlastMessage* var = message;
SBlastMessage* next = NULL;
while (var)
{
next = var->next;
sfree((var->message));
SeqIdSetFree(var->query_id);
var = next;
}
return NULL;
}
extern void RemoveInternalOrfs (SeqLocPtr PNTR slpFound)
{
SeqLocPtr slp1, slp2, slp = NULL;
SeqIdPtr id;
Int4 start1, stop1, start2, stop2;
Boolean flagInternal;
slp1 = *slpFound;
while (slp1 != NULL)
{
start1 = SeqLocStart (slp1);
stop1 = SeqLocStop (slp1);
flagInternal = FALSE;
slp2 = *slpFound;
while (slp2 != NULL)
{
start2 = SeqLocStart (slp2);
stop2 = SeqLocStop (slp2);
if ((start1 > start2 && start1 < stop2) &&
(stop1 > start2 && stop1 < stop2))
{
flagInternal = TRUE;
break;
}
slp2 = slp2->next;
}
if (!flagInternal)
SeqLocLink (&slp, SeqLocDup (slp1));
slp1 = slp1->next;
}
slp1 = *slpFound;
while (slp1 != NULL)
{
slp2 = slp1->next;
id = SeqLocId (slp1);
if (id != NULL)
id->next = SeqIdSetFree (id->next);
SeqLocFree (slp1);
slp1 = slp2;
}
*slpFound = slp;
return;
}
static void GetThisBioseq (XOSPtr xosp)
{
GatherScopePtr gsp;
CharPtr filename;
Int4 gi;
FILE *fiop;
Boolean flagHaveNet;
SeqEntryPtr sep;
ValNodePtr vnp;
Int2 gcode;
GeneticCodePtr gcp;
SeqLocPtr slp, slpn;
SeqIdPtr id;
gsp = xosp->gsp;
fiop = NULL;
filename = xosp->filename;
gi = xosp->gi;
xosp->sep = SeqEntryFree (xosp->sep);
if (gi > 0)
{
if (!EntrezInit ("cnsgnv", FALSE, &flagHaveNet))
{
ErrPostEx (SEV_ERROR, TOP_ERROR, 102,
"Entrez init failed");
ErrShow ();
return;
}
}
if (gi > 0)
{
sep = EntrezSeqEntryGet (gi, SEQENTRY_READ_NUC_PROT);
}
else if (filename != NULL)
{
if ((fiop = FileOpen (filename, "r")) == NULL)
{
ErrPostEx (SEV_ERROR, TOP_ERROR, 103,
"Failed to open FastA file");
ErrShow ();
return;
}
sep = FastaToSeqEntry (fiop, TRUE);
AddBioSourceToSeqEntry (sep);
}
else
{
sep = NULL;
}
if (sep == NULL)
{
ErrPostEx (SEV_ERROR, TOP_ERROR, 104,
"No SeqEntry");
ErrShow ();
}
else
{
CleanUpXOS (xosp);
xosp->sep = sep;
xosp->gi = gi;
xosp->filename = StringSave (filename);
GatherSeqEntry (sep, (Pointer) xosp, GetBioseq, (Pointer) gsp);
}
if (xosp->bsp != NULL)
{
vnp = xosp->bsp->descr;
gcode = 0;
while (vnp != NULL)
{
if (vnp->choice == Seq_descr_source)
{
gcode = BioSourceToGeneticCode ((BioSourcePtr) vnp->data.ptrvalue);
break;
}
vnp = vnp->next;
}
if (gcode == 0 && sep->choice == 2)
{
vnp = ((BioseqSetPtr) (sep->data.ptrvalue))->descr;
while (vnp != NULL)
{
if (vnp->choice == Seq_descr_source)
{
gcode = BioSourceToGeneticCode ((BioSourcePtr) vnp->data.ptrvalue);
break;
}
vnp = vnp->next;
}
}
if (gcode == 0)
gcode = 1; /* standard */
gcp = GeneticCodeFind (gcode, NULL);
if (gcp != NULL)
{
xosp->gcd = xosp->gcdi = NULL;
vnp = (ValNodePtr) gcp->data.ptrvalue;
while (vnp != NULL)
{
if (vnp->choice == 6) /* sncbieaa */
xosp->gcdi = (CharPtr) vnp->data.ptrvalue;
else if (vnp->choice == 3) /* ncbieaa */
xosp->gcd = (CharPtr) vnp->data.ptrvalue;
vnp = vnp->next;
}
}
if (xosp->gcdi == NULL)
xosp->gcdi = xosp->gcd;
if (xosp->gcdi == NULL)
{
ErrPostEx (SEV_ERROR, TOP_ERROR, 105,
"Could not get genetic code for translation");
ErrShow ();
xosp->bsp = NULL;
}
}
if (gi > 0)
{
slp = xosp->slpk;
while (slp != NULL)
{
slpn = slp->next;
id = SeqLocId (slp);
if (id != NULL)
id->next = SeqIdSetFree (id->next);
SeqLocFree (slp);
slp = slpn;
}
xosp->slpk = slp;
GatherSeqEntry (sep, (Pointer) xosp, GatherKnownOrfs, (Pointer) gsp);
}
if (gi > 0)
EntrezFini ();
else
FileClose (fiop);
if (xosp->bsp != NULL)
{
if (!ISA_na (xosp->bsp->mol))
{
xosp->sep = SeqEntryFree (xosp->sep);
xosp->bsp = NULL;
xosp->filename = (CharPtr) MemFree (xosp->filename);
xosp->gi = 0;
ErrPostEx (SEV_ERROR, TOP_ERROR, 101, "Not nucleic acid Bioseq");
ErrShow ();
}
}
else
{
xosp->sep = SeqEntryFree (xosp->sep);
xosp->filename = (CharPtr) MemFree (xosp->filename);
xosp->gi = 0;
ErrPostEx (SEV_ERROR, TOP_ERROR, 101, "No Bioseq");
ErrShow ();
}
return;
}
static int LIBCALLBACK
MegaBlastPrintEndpoints(VoidPtr ptr)
{
BlastSearchBlkPtr search = (BlastSearchBlkPtr) ptr;
CharPtr subject_descr;
SeqIdPtr sip, query_id;
CharPtr query_buffer, title;
CharPtr subject_buffer;
Int4 query_length, q_start, q_end, q_shift=0, s_shift=0;
Int4 subject_end;
Int4 hsp_index;
Boolean numeric_sip_type = FALSE;
BLAST_HSPPtr hsp;
Int2 context;
Char context_sign;
Int4 subject_gi, score;
FILE *fp = (FILE *) search->output;
if (search->current_hitlist == NULL || search->current_hitlist->hspcnt <= 0) {
search->subject_info = BLASTSubjectInfoDestruct(search->subject_info);
return 0;
}
if (search->rdfp)
readdb_get_descriptor(search->rdfp, search->subject_id, &sip,
&subject_descr);
else
sip = SeqIdSetDup(search->subject_info->sip);
if (sip->choice != SEQID_GENERAL ||
StringCmp(((DbtagPtr)sip->data.ptrvalue)->db, "BL_ORD_ID")) {
if (search->pbp->mb_params->full_seqids) {
subject_buffer = (CharPtr) Malloc(BUFFER_LENGTH + 1);
SeqIdWrite(sip, subject_buffer, PRINTID_FASTA_LONG, BUFFER_LENGTH);
} else
numeric_sip_type = GetAccessionFromSeqId(SeqIdFindBest(sip, SEQID_GI),
&subject_gi, &subject_buffer);
} else {
DbtagPtr db_tag = (DbtagPtr) sip->data.ptrvalue;
if (db_tag->db &&
(!StringCmp(db_tag->db, "THC") ||
!StringICmp(db_tag->db, "TI")) &&
db_tag->tag->id != 0) {
subject_buffer = (CharPtr) Malloc(16);
sprintf(subject_buffer, "%ld", (long) db_tag->tag->id);
} else {
subject_buffer = StringTokMT(subject_descr, " \t", &subject_descr);
subject_descr = subject_buffer;
}
}
search->current_hitlist->hspcnt_max = search->current_hitlist->hspcnt;
/* Only for the two sequences case, get offset shift if subject
is a subsequence */
if (!search->rdfp && search->query_slp->next) {
s_shift = SeqLocStart(search->query_slp->next);
subject_end = SeqLocStop(search->query_slp->next);
} else {
s_shift = 0;
subject_end =
readdb_get_sequence_length(search->rdfp, search->subject_id);
}
/* Get offset shift if query is a subsequence */
q_shift = SeqLocStart(search->query_slp);
for (hsp_index=0; hsp_index<search->current_hitlist->hspcnt; hsp_index++) {
hsp = search->current_hitlist->hsp_array[hsp_index];
if (hsp==NULL || (search->pbp->cutoff_e > 0 &&
hsp->evalue > search->pbp->cutoff_e))
continue;
/* Correct query context is already found in BlastGetNonSumStatsEvalue */
context = hsp->context;
query_id = search->qid_array[context/2];
if (query_id == NULL) /* Bad hsp, something wrong */
continue;
hsp->context = context & 1;
query_length = search->query_context_offsets[context+1] -
search->query_context_offsets[context] - 1;
hsp->subject.end = hsp->subject.offset + hsp->subject.length;
if (hsp->context) {
hsp->query.end = query_length - hsp->query.offset;
hsp->query.offset =
hsp->query.end - hsp->query.length + 1;
context_sign = '-';
} else {
hsp->query.end = (++hsp->query.offset) + hsp->query.length - 1;
if (hsp->query.end > query_length) {
hsp->subject.end -= (hsp->query.end - query_length);
hsp->query.end = query_length;
}
context_sign = '+';
}
if (hsp->subject.end > subject_end) {
hsp->query.end -= (hsp->subject.end - subject_end);
hsp->subject.end = subject_end;
}
hsp->subject.offset++;
query_buffer = NULL;
if (query_id->choice == SEQID_LOCAL &&
search->pbp->mb_params->full_seqids) {
BioseqPtr query_bsp = BioseqLockById(query_id);
title = StringSave(BioseqGetTitle(query_bsp));
if (title)
query_buffer = StringTokMT(title, " ", &title);
else {
Int4 query_gi;
GetAccessionFromSeqId(query_bsp->id, &query_gi,
&query_buffer);
}
BioseqUnlock(query_bsp);
} else {
query_buffer = (CharPtr) Malloc(BUFFER_LENGTH + 1);
if (!search->pbp->mb_params->full_seqids)
SeqIdWrite(query_id, query_buffer, PRINTID_TEXTID_ACCESSION,
BUFFER_LENGTH);
else
SeqIdWrite(query_id, query_buffer, PRINTID_FASTA_LONG,
BUFFER_LENGTH);
}
if (search->pbp->gap_open==0 && search->pbp->gap_extend==0)
score = ((hsp->subject.length + hsp->query.length)*
search->sbp->reward / 2 - hsp->score) /
(search->sbp->reward - search->sbp->penalty);
else
score = hsp->score;
if (context_sign == '+') {
q_start = hsp->query.offset;
q_end = hsp->query.end;
} else {
q_start = hsp->query.end;
q_end = hsp->query.offset;
}
/* Adjust offsets if query is a subsequence, only for first query */
if (context < 2) {
q_start += q_shift;
q_end += q_shift;
}
hsp->subject.offset += s_shift;
hsp->subject.end += s_shift;
if (numeric_sip_type)
fprintf(fp, "'%ld'=='%c%s' (%d %d %d %d) %d\n", (long) subject_gi,
context_sign, query_buffer, hsp->subject.offset, q_start,
hsp->subject.end, q_end, score);
else
fprintf(fp, "'%s'=='%c%s' (%d %d %d %d) %d\n",
subject_buffer, context_sign, query_buffer,
hsp->subject.offset, q_start,
hsp->subject.end, q_end, score);
MemFree(query_buffer);
}
if (!numeric_sip_type && subject_buffer != subject_descr)
MemFree(subject_buffer);
MemFree(subject_descr);
sip = SeqIdSetFree(sip);
return 0;
}
static int LIBCALLBACK
MegaBlastPrintSegments(VoidPtr ptr)
{
BlastSearchBlkPtr search = (BlastSearchBlkPtr) ptr;
ReadDBFILEPtr rdfp = search->rdfp;
BLAST_HSPPtr hsp;
Int4 i, subject_gi;
Int2 context;
CharPtr query_buffer, title;
SeqIdPtr sip, query_id;
Int4 hsp_index, score;
Uint1Ptr query_seq, subject_seq = NULL;
FloatHi perc_ident;
Char strand;
GapXEditScriptPtr esp;
Int4 q_start, q_end, s_start, s_end, query_length, numseg;
Int4 q_off, num_ident, align_length, total_ident, q_shift=0, s_shift=0;
Int4Ptr length, start;
Uint1Ptr strands;
CharPtr subject_descr, subject_buffer, buffer;
Char tmp_buffer[BUFFER_LENGTH];
Int4 buffer_size, max_buffer_size = LARGE_BUFFER_LENGTH;
Boolean numeric_sip_type = FALSE;
FILE *fp = (FILE *) search->output;
if (search->current_hitlist == NULL || search->current_hitlist->hspcnt <= 0) {
search->subject_info = BLASTSubjectInfoDestruct(search->subject_info);
return 0;
}
subject_seq = search->subject->sequence_start + 1;
if (rdfp)
readdb_get_descriptor(rdfp, search->subject_id, &sip, &subject_descr);
else
sip = SeqIdSetDup(search->subject_info->sip);
if (sip->choice != SEQID_GENERAL ||
StringCmp(((DbtagPtr)sip->data.ptrvalue)->db, "BL_ORD_ID")) {
if (search->pbp->mb_params->full_seqids) {
subject_buffer = (CharPtr) Malloc(BUFFER_LENGTH + 1);
SeqIdWrite(sip, subject_buffer, PRINTID_FASTA_LONG, BUFFER_LENGTH);
} else
numeric_sip_type = GetAccessionFromSeqId(SeqIdFindBest(sip, SEQID_GI),
&subject_gi, &subject_buffer);
} else {
subject_buffer = StringTokMT(subject_descr, " \t", &subject_descr);
subject_descr = subject_buffer;
}
buffer = (CharPtr) Malloc(LARGE_BUFFER_LENGTH);
/* Only for the two sequences case, get offset shift if subject
is a subsequence */
if (!rdfp && search->query_slp->next)
s_shift = SeqLocStart(search->query_slp->next);
/* Get offset shift if query is a subsequence */
q_shift = SeqLocStart(search->query_slp);
for (hsp_index=0; hsp_index<search->current_hitlist->hspcnt; hsp_index++) {
hsp = search->current_hitlist->hsp_array[hsp_index];
if (hsp==NULL || (search->pbp->cutoff_e > 0 &&
hsp->evalue > search->pbp->cutoff_e)) {
continue;
}
context = hsp->context;
query_id = search->qid_array[context/2];
if (query_id == NULL) /* Bad hsp, something wrong */
continue;
hsp->context = context & 1;
if (search->pbp->gap_open==0 && search->pbp->gap_extend==0)
score = ((hsp->subject.length + hsp->query.length)*
search->sbp->reward / 2 - hsp->score) /
(search->sbp->reward - search->sbp->penalty);
else
score = hsp->score;
query_length = search->query_context_offsets[context+1] -
search->query_context_offsets[context] - 1;
q_off = hsp->query.offset;
if (hsp->context) {
strand = '-';
hsp->query.end = query_length - hsp->query.offset;
hsp->query.offset =
hsp->query.end - hsp->query.length;
} else {
strand = '+';
hsp->query.end = hsp->query.offset + hsp->query.length;
}
if (strand == '+') {
q_start = hsp->query.offset + 1;
q_end = hsp->query.end;
} else {
q_start = hsp->query.end;
q_end = hsp->query.offset + 1;
}
s_start = hsp->subject.offset + 1;
s_end = hsp->subject.offset + hsp->subject.length;
/* Adjust offsets if query is a subsequence, only for first query */
if (context < 2) {
q_start += q_shift;
q_end += q_shift;
}
s_start += s_shift;
s_end += s_shift;
if (query_id->choice == SEQID_LOCAL &&
search->pbp->mb_params->full_seqids) {
BioseqPtr query_bsp = BioseqLockById(query_id);
title = StringSave(BioseqGetTitle(query_bsp));
if (title)
query_buffer = StringTokMT(title, " ", &title);
else {
Int4 query_gi;
GetAccessionFromSeqId(query_bsp->id, &query_gi,
&query_buffer);
}
BioseqUnlock(query_bsp);
} else {
query_buffer = (CharPtr) Malloc(BUFFER_LENGTH + 1);
if (!search->pbp->mb_params->full_seqids)
SeqIdWrite(query_id, query_buffer, PRINTID_TEXTID_ACCESSION,
BUFFER_LENGTH);
else
SeqIdWrite(query_id, query_buffer, PRINTID_FASTA_LONG,
BUFFER_LENGTH);
}
if (numeric_sip_type)
sprintf(buffer, "\n#'>%ld'=='%c%s' (%d %d %d %d) %d\na {\n s %d\n b %d %d\n e %d %d\n",
(long) subject_gi, strand, query_buffer,
s_start, q_start, s_end, q_end, score, score,
s_start, q_start, s_end, q_end);
else
sprintf(buffer, "\n#'>%s'=='%c%s' (%d %d %d %d) %d\na {\n s %d\n b %d %d\n e %d %d\n",
subject_buffer, strand, query_buffer,
s_start, q_start, s_end, q_end, score, score,
s_start, q_start, s_end, q_end);
buffer_size = StringLen(buffer);
query_seq = search->context[context].query->sequence;
esp = hsp->gap_info->esp;
for (numseg=0; esp; esp = esp->next, numseg++);
GXECollectDataForSeqalign(hsp->gap_info, hsp->gap_info->esp, numseg,
&start, &length, &strands,
&q_off, &hsp->subject.offset);
if (start[0] < 0) {
length[0] += start[0];
start[1] -= start[0];
start[0] = 0;
}
if (start[2*(numseg-1)] + length[numseg-1] > query_length)
length[numseg-1] = query_length - start[2*(numseg-1)];
total_ident = 0;
align_length = 0;
for (i=0; i<numseg; i++) {
align_length += length[i];
if (strand == '+') {
q_start = start[2*i] + 1;
q_end = q_start + length[i] - 1;
} else {
q_start = query_length - start[2*i];
q_end = q_start - length[i] + 1;
}
if (start[2*i] != -1 && start[2*i+1] != -1) {
num_ident = MegaBlastGetNumIdentical(query_seq, subject_seq,
start[2*i], start[2*i+1],
length[i], FALSE);
perc_ident = (FloatHi) num_ident / length[i] * 100;
total_ident += num_ident;
sprintf(tmp_buffer, " l %d %d %d %d (%.0f)\n", start[2*i+1]+1,
q_start, start[2*i+1]+length[i],
q_end, perc_ident);
if ((buffer_size += StringLen(tmp_buffer)) > max_buffer_size - 2) {
max_buffer_size *= 2;
buffer = (CharPtr) Realloc(buffer, max_buffer_size);
}
StringCat(buffer, tmp_buffer);
}
}
if (100*total_ident >=
align_length*search->pbp->mb_params->perc_identity) {
StringCat(buffer, "}");
fprintf(fp, "%s\n", buffer);
}
MemFree(start);
MemFree(length);
MemFree(strands);
MemFree(query_buffer);
} /* End loop on hsp's */
if (!numeric_sip_type && subject_buffer != subject_descr)
MemFree(subject_buffer);
MemFree(subject_descr);
MemFree(buffer);
sip = SeqIdSetFree(sip);
fflush(fp);
return 1;
}
extern Gather_CDSPtr GatherCDSFree (Gather_CDSPtr gcdsp)
{
SeqLocPtr slp, slpn;
SeqIdPtr id;
if (gcdsp == NULL)
return NULL;
MemFree (gcdsp->tableGlobal);
MemFree (gcdsp->tableRefine);
slp = gcdsp->slpGlobal;
while (slp != NULL)
{
slpn = slp->next;
id = SeqLocId (slp);
if (id != NULL)
id->next = SeqIdSetFree (id->next);
SeqLocFree (slp);
slp = slpn;
}
slp = gcdsp->slpRefine;
while (slp != NULL)
{
slpn = slp->next;
id = SeqLocId (slp);
if (id != NULL)
id->next = SeqIdSetFree (id->next);
SeqLocFree (slp);
slp = slpn;
}
slp = gcdsp->slpAll;
while (slp != NULL)
{
slpn = slp->next;
id = SeqLocId (slp);
if (id != NULL)
id->next = SeqIdSetFree (id->next);
SeqLocFree (slp);
slp = slpn;
}
slp = gcdsp->slpHit;
while (slp != NULL)
{
slpn = slp->next;
id = SeqLocId (slp);
if (id != NULL)
id->next = SeqIdSetFree (id->next);
SeqLocFree (slp);
slp = slpn;
}
slp = gcdsp->slpFound;
while (slp != NULL)
{
slpn = slp->next;
id = SeqLocId (slp);
if (id != NULL)
id->next = SeqIdSetFree (id->next);
SeqLocFree (slp);
slp = slpn;
}
return (Gather_CDSPtr) MemFree (gcdsp);
}