void
Blast_SeqIdGetDefLine(SeqId* sip, char** buffer_ptr, Boolean ncbi_gi,
Boolean accession_only, Boolean search_for_id)
{
char* seqid_buffer = NULL;
Int4 gi = 0;
Boolean numeric_id_type = FALSE;
*buffer_ptr = NULL;
if (sip == NULL)
return;
/* Check for ad hoc ID's generated by formatdb if the user does not provide
any. */
if (search_for_id && (sip->choice != SEQID_GENERAL ||
StringCmp(((Dbtag*)sip->data.ptrvalue)->db, "BL_ORD_ID")))
{
if ((!accession_only && !ncbi_gi) || sip->choice == SEQID_LOCAL) {
seqid_buffer = (char*) malloc(BUFFER_LENGTH + 1);
SeqIdWrite(sip, seqid_buffer, PRINTID_FASTA_LONG, BUFFER_LENGTH);
} else if (accession_only) {
seqid_buffer = (char*) malloc(BUFFER_LENGTH + 1);
SeqIdWrite(SeqIdFindBestAccession(sip), seqid_buffer,
PRINTID_TEXTID_ACC_VER, BUFFER_LENGTH);
} else if (ncbi_gi) {
numeric_id_type =
GetAccessionFromSeqId(SeqIdFindBest(sip, SEQID_GI),
&gi, &seqid_buffer);
} else {
numeric_id_type =
GetAccessionFromSeqId(SeqIdFindBestAccession(sip),
&gi, &seqid_buffer);
}
}
if (numeric_id_type && gi > 0) {
seqid_buffer = (char*) malloc(16);
sprintf(seqid_buffer, "%ld", (long) gi);
}
if (!seqid_buffer) {
/* If it's still NULL make a last ditch effort to get info. */
char* title=NULL;
Bioseq* bsp = BioseqLockById(sip);
if (bsp) {
if (BioseqGetTitle(bsp) != NULL)
title = strdup(BioseqGetTitle(bsp));
else
title = strdup("No definition line found");
}
BioseqUnlock(bsp);
if (title) /* Use first token as id. */
seqid_buffer = StringTokMT(title, " \t\n\r", &title);
}
*buffer_ptr = seqid_buffer;
}
static ValNodePtr CollectBioseqLineValues (BioseqPtr bsp, ValNodePtr field_list, Boolean want_gi)
{
SeqDescrPtr sdp;
SeqMgrDescContext dcontext;
Char id_txt[255], id_txt2[255];
SeqIdPtr sip, sip_gi = NULL, sip_gb = NULL;
ValNodePtr line_list = NULL, line_values;
if (bsp == NULL) {
return NULL;
}
for (sip = bsp->id; sip != NULL; sip = sip->next) {
if (sip->choice == SEQID_GENBANK
|| (sip->choice == SEQID_EMBL && sip_gb == NULL)
|| (sip->choice == SEQID_SWISSPROT && sip_gb == NULL)
|| (sip->choice == SEQID_DDBJ && sip_gb == NULL)
|| (sip->choice == SEQID_PIR && sip_gb == NULL)) {
sip_gb = sip;
} else if (sip->choice == SEQID_GI) {
sip_gi = sip;
}
}
if (sip_gb == NULL && sip_gi == NULL) {
SeqIdWrite (SeqIdFindBest (bsp->id, SEQID_GENBANK), id_txt, PRINTID_REPORT, sizeof (id_txt) - 1);
id_txt2[0] = 0;
} else {
if (sip_gb == NULL) {
id_txt[0] = 0;
} else {
SeqIdWrite (sip_gb, id_txt, PRINTID_REPORT, sizeof (id_txt) - 1);
}
if (sip_gi == NULL) {
id_txt2[0] = 0;
} else {
SeqIdWrite (sip_gi, id_txt2, PRINTID_REPORT, sizeof (id_txt2) - 1);
}
}
for (sdp = SeqMgrGetNextDescriptor (bsp, NULL, Seq_descr_source, &dcontext);
sdp != NULL;
sdp = SeqMgrGetNextDescriptor (bsp, sdp, Seq_descr_source, &dcontext)) {
line_values = NULL;
ValNodeAddPointer (&line_values, 0, StringSave (id_txt));
if (want_gi) {
ValNodeAddPointer (&line_values, 0, StringSave (id_txt2));
}
ValNodeLink (&line_values, CollectBioSourceValues (sdp->data.ptrvalue, field_list));
ValNodeAddPointer (&line_list, 0, line_values);
}
return line_list;
}
static void PrintBioseqLines (FILE *fp, BioseqPtr bsp, ValNodePtr field_list, Boolean want_gi)
{
SeqDescrPtr sdp;
SeqMgrDescContext dcontext;
Char id_txt[255], id_txt2[255];
SeqIdPtr sip, sip_gi = NULL, sip_gb = NULL;
if (fp == NULL || bsp == NULL) {
return;
}
for (sip = bsp->id; sip != NULL; sip = sip->next) {
if (sip->choice == SEQID_GENBANK
|| (sip->choice == SEQID_EMBL && sip_gb == NULL)
|| (sip->choice == SEQID_SWISSPROT && sip_gb == NULL)
|| (sip->choice == SEQID_DDBJ && sip_gb == NULL)
|| (sip->choice == SEQID_PIR && sip_gb == NULL)) {
sip_gb = sip;
} else if (sip->choice == SEQID_GI) {
sip_gi = sip;
}
}
if (sip_gb == NULL && sip_gi == NULL) {
SeqIdWrite (SeqIdFindBest (bsp->id, SEQID_GENBANK), id_txt, PRINTID_REPORT, sizeof (id_txt) - 1);
id_txt2[0] = 0;
} else {
if (sip_gb == NULL) {
id_txt[0] = 0;
} else {
SeqIdWrite (sip_gb, id_txt, PRINTID_REPORT, sizeof (id_txt) - 1);
}
if (sip_gi == NULL) {
id_txt2[0] = 0;
} else {
SeqIdWrite (sip_gi, id_txt2, PRINTID_REPORT, sizeof (id_txt2) - 1);
}
}
for (sdp = SeqMgrGetNextDescriptor (bsp, NULL, Seq_descr_source, &dcontext);
sdp != NULL;
sdp = SeqMgrGetNextDescriptor (bsp, sdp, Seq_descr_source, &dcontext)) {
if (want_gi) {
fprintf (fp, "%s\t%s", id_txt, id_txt2);
} else {
fprintf (fp, "%s", id_txt);
}
PrintBioSourceLine (fp, sdp->data.ptrvalue, field_list);
fprintf (fp, "\n");
}
}
static Boolean UpdateContigReadId (TContigReadPtr read, SeqIdReplaceListPtr pair_list, Boolean no_lookup, Boolean is_srr, char *has_errors)
{
SeqIdPairPtr pair;
SeqIdPtr sip_find;
Char id_buf[255];
Boolean rval = TRUE;
if (read == NULL || StringHasNoText (read->read_id)) {
rval = FALSE;
} else {
sip_find = MakeSeqID (read->read_id);
pair = FindReplacementInSeqIdReplaceList (sip_find, pair_list);
if (pair != NULL && (no_lookup || OkToReplaceId (pair, read->read_seq, has_errors))) {
if (pair->is_complement) {
if (read->is_complement) {
read->is_complement = FALSE;
} else {
read->is_complement = TRUE;
}
}
if (pair->ti > 0) {
read->ti = pair->ti;
} else {
if (pair->sip_replace->choice == SEQID_LOCAL) {
SeqIdWrite (pair->sip_replace, id_buf, PRINTID_REPORT, sizeof (id_buf) - 1);
} else {
SeqIdWrite (pair->sip_replace, id_buf, PRINTID_FASTA_LONG, sizeof (id_buf) - 1);
}
if (is_srr) {
if (read->srr != NULL) {
free (read->srr);
}
read->srr = malloc (sizeof (Char) * (StringLen (id_buf) + 1));
sprintf (read->srr, "%s", id_buf);
free (read->read_id);
read->read_id = NULL;
} else {
free (read->read_id);
read->read_id = malloc (sizeof (Char) * (StringLen (id_buf) + 1));
sprintf (read->read_id, "%s", id_buf);
}
}
read->local = FALSE;
}
sip_find = SeqIdFree (sip_find);
}
return rval;
}
static SeqIdPairPtr FindReplacementInSeqIdReplaceList (SeqIdPtr sip, SeqIdReplaceListPtr pair_list)
{
Int4 l, r, m;
Char buf_find[100];
int cmp;
if (sip == NULL || pair_list == NULL) return NULL;
SeqIdWrite (sip, buf_find, PRINTID_REPORT, sizeof (buf_find) - 1);
l = 0;
r = pair_list->num_ids - 1;
m = (r + l) / 2;
while ((cmp = StringICmp (buf_find, pair_list->list[m].buf_find)) != 0 && l <= r) {
if (cmp < 0) {
r = m - 1;
} else {
l = m + 1;
}
m = (r + l) / 2;
}
if (cmp == 0) {
return pair_list->list + m;
} else {
return NULL;
}
}
/** Fills the Iteration ASN.1 structure, for part of the BLAST XML report
* corresponding to one query.
* @param seqalign Seq-align list with results [in]
* @param sum_returns Search summary data [in]
* @param is_ooframe Was out-of-frame gapping used in this search? [in]
* @param ungapped Was this an ungapped search? [in]
* @param iter_num Index of this "iteration" (query). [in]
* @param message Error or warning message [in]
* @param query Query Bioseq [in]
* @param mask_loc List of masking locations [in]
* @return Populated structure.
*/
static Iteration*
s_XMLBuildOneQueryIteration(SeqAlign* seqalign,
Blast_SummaryReturn* sum_returns,
Boolean is_ooframe, Boolean ungapped,
Int4 iter_num, char* message,
Bioseq* query, ValNode* mask_loc)
{
Iteration* iterp = IterationNew();
iterp->iter_num = iter_num;
if (query) {
char buffer[1024];
SeqIdWrite(query->id, buffer, PRINTID_FASTA_LONG, sizeof(buffer));
iterp->query_ID = strdup(buffer);
if(BioseqGetTitle(query) != NULL)
iterp->query_def = strdup(BioseqGetTitle(query));
else
iterp->query_def = strdup("No definition line found");
iterp->query_len = query->length;
}
if(seqalign != NULL) {
iterp->hits =
BXMLSeqAlignToHits(seqalign, ungapped, is_ooframe, mask_loc);
}
iterp->stat = s_XMLBuildStatistics(sum_returns, ungapped);
if (message)
iterp->message = strdup(message);
return iterp;
}
static void ReportInvalidReplacement (SeqIdPtr sip, CharPtr reason, char *has_errors)
{
Char buf[128];
SeqIdWrite (sip, buf, PRINTID_FASTA_LONG, sizeof (buf) - 1);
PrintACEFormatErrorXMLStart (buf, has_errors);
printf ("%s", reason);
PrintACEFormatErrorXMLEnd ();
}
static void DoSuggestIntervals (
BioseqPtr bsp,
Pointer userdata
)
{
CharPtr caret5, caret3;
CSpeedFlagPtr cfp;
Char id [64];
SeqLocPtr loc, slp;
Boolean partial5, partial3;
SeqAnnotPtr sap;
SeqFeatPtr sfp;
SeqIdPtr sip;
Int4 start, stop;
if (bsp == NULL) return;
if (! ISA_aa (bsp->mol)) return;
cfp = (CSpeedFlagPtr) userdata;
if (cfp == NULL || cfp->ofp == NULL || cfp->nucbsp == NULL) return;
sip = SeqIdFindBest (bsp->id, 0);
if (sip == NULL) return;
SeqIdWrite (sip, id, PRINTID_FASTA_SHORT, sizeof (id) - 1);
sap = SuggestCodingRegion (cfp->nucbsp, bsp, cfp->genCode);
if (sap == NULL) return;
if (sap->type == 1) {
sfp = (SeqFeatPtr) sap->data;
if (sfp != NULL && sfp->data.choice == SEQFEAT_CDREGION) {
loc = sfp->location;
if (loc != NULL) {
fprintf (cfp->ofp, "%s\n", id);
slp = SeqLocFindNext (loc, NULL);
while (slp != NULL) {
start = GetOffsetInBioseq (slp, cfp->nucbsp, SEQLOC_START) + 1;
stop = GetOffsetInBioseq (slp, cfp->nucbsp, SEQLOC_STOP) + 1;
caret5 = "";
caret3 = "";
CheckSeqLocForPartial (slp, &partial5, &partial3);
if (partial5) {
caret5 = "<";
}
if (partial3) {
caret3 = ">";
}
fprintf (cfp->ofp, "%s%ld\t%s%ld\n", caret5, (long) start, caret3, (long) stop);
slp = SeqLocFindNext (loc, slp);
}
}
}
}
SeqAnnotFree (sap);
}
static void VecScreen_AddToQueue (
VQUEUE * queue,
VecScreenResultProc resultproc,
VecScreenAnnounceProc announceproc,
Nlm_VoidPtr userdata,
CONN conn,
BioseqPtr bsp
)
{
VQueuePtr cqp;
VQueuePtr PNTR qptr;
VQueuePtr tmp;
if (queue == NULL || resultproc == NULL || conn == NULL || bsp == NULL) return;
/* allocate queue element */
cqp = (VQueuePtr) MemNew (sizeof (VecScreenQueue));
if (cqp == NULL) return;
cqp->rid [0] = '\0';
SeqIdWrite (bsp->id, cqp->seqid, PRINTID_FASTA_LONG, 40);
cqp->estTime = 0;
cqp->initialTime = GetSecs ();
cqp->postedTime = cqp->initialTime;
cqp->secondsToWait = 15;
cqp->resultproc = resultproc;
cqp->announceproc = announceproc;
cqp->userdata = userdata;
cqp->connqueue = NULL;
cqp->done = FALSE;
/* add to polling queue */
qptr = (VQueuePtr PNTR) queue;
if (qptr != NULL) {
if (*qptr != NULL) {
tmp = *qptr;
if (tmp != NULL) {
while (tmp->next != NULL) {
tmp = tmp->next;
}
tmp->next = cqp;
}
} else {
*qptr = cqp;
}
}
/* queue the request for a rID */
QUERY_AddToQueue (&(cqp->connqueue), conn, FirstVecScreenCallback, (Pointer) cqp, TRUE);
}
extern CharPtr ErrorDescString (SeqIdPtr sip)
{
SeqIdPtr bestid;
CharPtr errbuf;
bestid = SeqIdFindBest(sip, SEQID_GI);
errbuf = (CharPtr) MemNew ((size_t) (sizeof (Char) * 32));
SeqIdWrite (bestid, errbuf, PRINTID_FASTA_LONG, 32-1);
return errbuf;
}
NLM_EXTERN SeqIdReplaceListPtr ReadSeqIdPairListFromFile (FILE *fp)
{
ReadBufferData rbd;
CharPtr linestring, cp, id2, buf = NULL;
Int4 len, buf_len = 0;
SeqIdPairPtr pair;
ValNodePtr pair_list = NULL, last = NULL, vnp;
SeqIdReplaceListPtr replace_list = NULL;
if (fp == NULL) return NULL;
rbd.fp = fp;
rbd.current_data = NULL;
linestring = AbstractReadFunction (&rbd);
while (linestring != NULL && linestring[0] != EOF) {
cp = linestring + StringSpn (linestring, " \t");
if (*cp != 0) {
len = StringCSpn (cp, " \t");
id2 = cp + len + StringSpn (cp + len, " \t");
if (*id2 != 0) {
if (len + 1 > buf_len) {
buf = MemFree (buf);
buf_len = len + 1;
buf = (CharPtr) MemNew (sizeof (Char) * buf_len);
}
StringNCpy (buf, cp, len);
buf[len] = 0;
pair = SeqIdPairNew ();
pair->sip_find = MakeSeqID (buf);
SeqIdWrite (pair->sip_find, pair->buf_find, PRINTID_REPORT, sizeof (pair->buf_find) - 1);
pair->sip_replace = MakeSeqID (id2);
vnp = ValNodeNew (NULL);
vnp->data.ptrvalue = pair;
if (last == NULL) {
pair_list = vnp;
} else {
last->next = vnp;
}
last = vnp;
}
}
free (linestring);
linestring = AbstractReadFunction (&rbd);
}
pair_list = ValNodeSort (pair_list, SortSeqIdPairList);
replace_list = SeqIdReplaceListNew (pair_list);
pair_list = SeqIdPairListFree (pair_list);
return replace_list;
}
static void DoVisitCodingRegions (
SeqFeatPtr sfp,
Pointer userdata
)
{
BioseqPtr bsp;
CharPtr caret5, caret3;
CSpeedFlagPtr cfp;
Char id [64];
SeqLocPtr loc, slp;
Boolean partial5, partial3;
SeqIdPtr sip;
Int4 start, stop;
if (sfp == NULL || sfp->data.choice != SEQFEAT_CDREGION) return;
cfp = (CSpeedFlagPtr) userdata;
if (cfp == NULL || cfp->ofp == NULL) return;
loc = sfp->location;
bsp = BioseqFindFromSeqLoc (loc);
if (bsp == NULL) return;
StringCpy (id, "?");
if (sfp->product != NULL) {
sip = SeqLocId (sfp->product);
if (sip != NULL) {
SeqIdWrite (sip, id, PRINTID_FASTA_SHORT, sizeof (id) - 1);
}
}
fprintf (cfp->ofp, "%s\n", id);
slp = SeqLocFindNext (loc, NULL);
while (slp != NULL) {
start = GetOffsetInBioseq (slp, bsp, SEQLOC_START) + 1;
stop = GetOffsetInBioseq (slp, bsp, SEQLOC_STOP) + 1;
caret5 = "";
caret3 = "";
CheckSeqLocForPartial (slp, &partial5, &partial3);
if (partial5) {
caret5 = "<";
}
if (partial3) {
caret3 = ">";
}
fprintf (cfp->ofp, "%s%ld\t%s%ld\n", caret5, (long) start, caret3, (long) stop);
slp = SeqLocFindNext (loc, slp);
}
}
static void PrintGraphMessage (BioseqPtr bsp, ScanDataPtr sdp,
CharPtr prefix, CharPtr suffix)
{
Char buf [41];
if (bsp == NULL || sdp == NULL || prefix == NULL) return;
SeqIdWrite (bsp->id, buf, PRINTID_FASTA_LONG, sizeof (buf));
fprintf (sdp->fp, "QA - %s - %s", buf, prefix);
if (! StringHasNoText (suffix)) {
fprintf (sdp->fp, " - %s", suffix);
}
fprintf (sdp->fp, "\n");
}
static void PrintBioseqErrorLine (FILE *fp, SeqIdPtr sip)
{
Char id_txt[255];
if (fp == NULL || sip == NULL) {
return;
}
SeqIdWrite (sip, id_txt, PRINTID_REPORT, sizeof (id_txt) - 1);
if (sip->choice == SEQID_GI) {
fprintf (fp, "\t%s\n", id_txt);
} else {
fprintf (fp, "%s\t\n", id_txt);
}
}
/*******************************************************************************
Function : DDV_DrawSequenceName()
Purpose : draw the name of the sequence (left column of the DDV panel)
Parameters : GrData; graphical data (font size, etc)
ScaleStyle;style of the ParaG scale
top, left; coord to start the draw
Return value : none
*******************************************************************************/
static void DDV_DrawSequenceName(UnDViewerGraphDataPtr GrData,ParaGPtr pgp,
Int2 top,Int2 left,Int4 cur_row,Int4 CurEditRow,Int4 CurMasterRow)
{
SeqIdPtr sip = NULL;
RecT rc;
Int2 x,y,decal=1,size;/*text position/size*/
Char szAccess[21];
BioseqPtr bsp;
/*get a name*/
bsp = BioseqLockById(pgp->sip);
if(bsp) {
sip = SeqIdFindBestAccession(bsp->id);
BioseqUnlock(bsp);
}
if (!sip)
sip = SeqIdFindBest(pgp->sip, 0);
SeqIdWrite(sip, szAccess,PRINTID_TEXTID_ACCESSION, 20);
/*compute position*/
if (pgp->ScaleStyle==SCALE_POS_TOP) decal++;
/*draw name*/
size=StringWidth(szAccess);
x=left/*-GrData->udv_scale.cxLeftScale*/-size;
y=top+decal*GrData->udv_font.LineHeight;
MoveTo(x,y);
if (cur_row==CurEditRow){
Magenta();
}
PaintString (szAccess);
if (cur_row==CurMasterRow){
Blue();
MoveTo(x,y);
LineTo(x+size,y);
}
/*draw a little box (for selection a full sequence)*/
left+=GrData->udv_font.cxChar;
top+=GrData->udv_font.cxChar/2;
LoadRect(&rc,left,top,left+GrData->udv_font.cxChar,
top+GrData->udv_font.cxChar);
Blue();
PaintOval(&rc);
Black();
}
static void DoNewFastaDefline (
BioseqPtr bsp,
Pointer userdata
)
{
BioseqSetPtr bssp;
CSpeedFlagPtr cfp;
Char id [128];
CharPtr title;
if (bsp == NULL) return;
cfp = (CSpeedFlagPtr) userdata;
if (cfp == NULL) return;
if (StringChr (cfp->skip, 's') != NULL) {
if (bsp->idx.parenttype == OBJ_BIOSEQSET) {
bssp = (BioseqSetPtr) bsp->idx.parentptr;
if (bssp != NULL) {
if (bssp->_class == BioseqseqSet_class_segset ||
bssp->_class == BioseqseqSet_class_parts) return;
}
}
}
if (StringChr (cfp->skip, 'v') != NULL) {
if (bsp->repr == Seq_repr_virtual) return;
}
id [0] = '\0';
SeqIdWrite (bsp->id, id, PRINTID_FASTA_LONG, sizeof (id) - 1);
title = NewCreateDefLine (NULL, bsp, FALSE, FALSE);
if (StringHasNoText (title)) {
title = StringSave ("?");
}
if (cfp->ofp != NULL) {
fprintf (cfp->ofp, ">%s %s\n", id, title);
}
MemFree (title);
}
static void WriteOneProteinWithProduct (BioseqPtr bsp, Pointer data)
{
FastaExportOptionsPtr fe;
SeqFeatPtr prot;
SeqMgrFeatContext fcontext;
Char id [128];
if (bsp != NULL && ISA_aa (bsp->mol) && (fe = (FastaExportOptionsPtr) data) != NULL) {
prot = SeqMgrGetNextFeature (bsp, NULL, 0, FEATDEF_PROT, &fcontext);
if (prot == NULL) {
BioseqFastaStreamEx (bsp, fe->fp, fe->flags, fe->linelen, fe->blocklen, fe->grouplen,
TRUE, FALSE, FALSE);
} else {
SeqIdWrite (bsp->id, id, PRINTID_FASTA_LONG, sizeof (id) - 1);
fprintf (fe->fp, ">%s [prot=%s]\n", id, fcontext.label);
BioseqFastaStreamEx (bsp, fe->fp, fe->flags, fe->linelen, fe->blocklen, fe->grouplen,
FALSE, FALSE, FALSE);
}
}
}
static void PrintFeatureMessage (SeqFeatPtr sfp, ScanDataPtr sdp,
CharPtr prefix, CharPtr suffix)
{
BioseqPtr bsp;
Char buf [41];
if (sfp == NULL || sdp == NULL || prefix == NULL) return;
bsp = BioseqFindFromSeqLoc (sfp->location);
if (bsp != NULL) {
SeqIdWrite (bsp->id, buf, PRINTID_FASTA_LONG, sizeof (buf));
fprintf (sdp->fp, "%s - %s", prefix, buf);
} else {
fprintf (sdp->fp, "%s - %s", prefix, sdp->buf);
}
if (! StringHasNoText (suffix)) {
fprintf (sdp->fp, " - %s", suffix);
}
fprintf (sdp->fp, "\n");
}
static void DoFastaComp (
BioseqPtr bsp,
Pointer userdata,
Boolean ignoreExisting
)
{
Char buf [4096];
CSpeedFlagPtr cfp;
Char id [128];
CharPtr title;
if (bsp == NULL) return;
cfp = (CSpeedFlagPtr) userdata;
if (cfp == NULL) return;
id [0] = '\0';
SeqIdWrite (bsp->id, id, PRINTID_FASTA_LONG, sizeof (id) - 1);
buf [0] = '\0';
CreateDefLineExEx (NULL, bsp, buf, sizeof (buf) - 1, 0,
NULL, NULL, ignoreExisting, FALSE);
title = NewCreateDefLine (NULL, bsp, ignoreExisting, FALSE);
if (StringHasNoText (title)) {
title = StringSave ("?");
}
if (StringCmp (buf, title) != 0) {
if (cfp->ofp != NULL) {
fprintf (cfp->ofp, "< %s %s\n", id, buf);
fprintf (cfp->ofp, "> %s %s\n", id, title);
}
printf ("< %s %s\n", id, buf);
printf ("> %s %s\n", id, title);
fflush (stdout);
}
MemFree (title);
}
static Boolean ValidateContigReadId (TContigReadPtr read, char *has_errors)
{
SeqIdPairData pair;
Char id_buf[255];
Boolean rval = TRUE;
if (read == NULL || StringHasNoText (read->read_id)) {
rval = FALSE;
} else if (!read->local) {
rval = TRUE;
} else {
pair.sip_find = NULL;
pair.is_complement = FALSE;
pair.is_consensus = FALSE;
pair.trim3 = 0;
pair.trim5 = 0;
pair.sip_replace = MakeSeqID (read->read_id);
pair.ti = 0;
if (OkToReplaceId (&pair, read->read_seq, has_errors)) {
if (pair.is_complement && !read->is_complement) {
read->is_complement = TRUE;
} else if (!pair.is_complement && read->is_complement) {
read->is_complement = FALSE;
}
if (pair.ti > 0) {
read->ti = pair.ti;
} else {
SeqIdWrite (pair.sip_replace, id_buf, PRINTID_FASTA_LONG, sizeof (id_buf) - 1);
free (read->read_id);
read->read_id = malloc (sizeof (Char) * (StringLen (id_buf) + 1));
sprintf (read->read_id, "%s", id_buf);
}
read->local = FALSE;
}
pair.sip_replace = SeqIdFree (pair.sip_replace);
}
return rval;
}
static void DoFastaDefline (
BioseqPtr bsp,
Pointer userdata
)
{
Char buf [4096];
CSpeedFlagPtr cfp;
Char id [128];
if (bsp == NULL) return;
cfp = (CSpeedFlagPtr) userdata;
if (cfp == NULL) return;
id [0] = '\0';
SeqIdWrite (bsp->id, id, PRINTID_FASTA_LONG, sizeof (id) - 1);
buf [0] = '\0';
CreateDefLine (NULL, bsp, buf, sizeof (buf) - 1, 0, NULL, NULL);
if (cfp->ofp != NULL) {
fprintf (cfp->ofp, ">%s %s\n", id, buf);
}
}
NLM_EXTERN Boolean UpdateContigIds (TContigPtr contig, SeqIdReplaceListPtr pair_list, Boolean no_lookup, Boolean is_srr, char *has_errors)
{
Int4 i;
SeqIdPairPtr pair;
SeqIdPtr sip_find;
Char id_buf[255];
Boolean rval = TRUE;
if (contig == NULL) return FALSE;
if (pair_list == NULL) return TRUE;
if (contig->consensus_id != NULL) {
sip_find = MakeSeqID (contig->consensus_id);
pair = FindReplacementInSeqIdReplaceList (sip_find, pair_list);
if (pair != NULL && (no_lookup || OkToReplaceId (pair, contig->consensus_seq, has_errors))) {
if (pair->is_complement) {
if (contig->is_complement) {
contig->is_complement = FALSE;
} else {
contig->is_complement = TRUE;
}
}
SeqIdWrite (pair->sip_replace, id_buf, PRINTID_FASTA_LONG, sizeof (id_buf) - 1);
free (contig->consensus_id);
contig->consensus_id = malloc (sizeof (Char) * (StringLen (id_buf) + 1));
sprintf (contig->consensus_id, "%s", id_buf);
} else {
rval = FALSE;
}
sip_find = SeqIdFree (sip_find);
}
for (i = 0; i < contig->num_reads; i++) {
rval &= UpdateContigReadId (contig->reads[i], pair_list, no_lookup, is_srr, has_errors);
}
return rval;
}
static void DoRecord (SeqEntryPtr sep, Pointer userdata)
{
BioseqPtr bsp;
SeqEntryPtr nsep;
ScanDataPtr sdp;
sdp = (ScanDataPtr) userdata;
(sdp->recordCount)++;
nsep = FindNthBioseq (sep, 1);
if (nsep != NULL && IS_Bioseq (nsep)) {
bsp = (BioseqPtr) nsep->data.ptrvalue;
if (bsp != NULL) {
SeqIdWrite (bsp->id, sdp->buf, PRINTID_FASTA_LONG, sizeof (sdp->buf));
}
}
#ifdef OS_UNIX
/* printf ("%s\n", sdp->buf); */
#endif
VisitPubdescsInSep (sep, (Pointer) sdp, DoThesis);
/* check for 'genomic DNA' in DoTitle suppressed for bulk submissions */
sdp->bulk = FALSE;
VisitDescriptorsInSep (sep, (Pointer) sdp, LookForBulk);
VisitDescriptorsInSep (sep, (Pointer) sdp, DoTitle);
VisitFeaturesInSep (sep, (Pointer) sdp, DoImpCDSandTrna);
/* index for pseudo cds, impfeat peptides codon frame */
SeqMgrIndexFeatures (0, sep->data.ptrvalue);
VisitFeaturesInSep (sep, (Pointer) sdp, DoPseudoCDS);
VisitFeaturesInSep (sep, (Pointer) sdp, DoPeptide);
/* now cleanup, index for overlapping peptides */
SeriousSeqEntryCleanup (sep, NULL, NULL);
SeqMgrIndexFeatures (0, sep->data.ptrvalue);
VisitBioseqsInSep (sep, (Pointer) sdp, DoProteins);
/*
VisitBioseqsInSep (sep, (Pointer) sdp, DoGraphs);
*/
#if 0
{
Boolean hasUser = FALSE;
VisitFeaturesInSep (sep, (Pointer) &hasUser, DoUser);
if (hasUser && sdp->aop != NULL && sdp->atp_se != NULL) {
SeqEntryAsnWrite (sep, sdp->aop, sdp->atp_se);
}
}
#endif
}
static void DoProcess (
SeqEntryPtr sep,
Uint2 entityID,
CSpeedFlagPtr cfp
)
{
Char id [64];
ErrSev oldErrSev;
ValidStructPtr vsp;
if (sep == NULL || cfp == NULL) return;
if (StringChr (cfp->clean, 't') != NULL) {
VisitDescriptorsInSep (sep, NULL, MarkTitles);
DeleteMarkedObjects (entityID, 0, NULL);
}
if (StringChr (cfp->clean, 'a') != NULL) {
AssignIDsInEntity (entityID, 0, NULL);
}
if (StringChr (cfp->clean, 'b') != NULL) {
BasicSeqEntryCleanup (sep);
}
if (StringChr (cfp->clean, 's') != NULL) {
SeriousSeqEntryCleanup (sep, NULL, NULL);
}
if (StringChr (cfp->index, 'f') != NULL) {
SeqMgrIndexFeatures (entityID, 0);
}
if (StringChr (cfp->seq, 'c') != NULL) {
VisitBioseqsInSep (sep, (Pointer) cfp, DoFastaExist);
}
if (StringChr (cfp->seq, 'C') != NULL) {
VisitBioseqsInSep (sep, (Pointer) cfp, DoFastaRegen);
}
if (StringChr (cfp->seq, 's') != NULL) {
VisitBioseqsInSep (sep, (Pointer) cfp, DoFastaSeq);
}
if (StringChr (cfp->seq, 'S') != NULL) {
if (SeqMgrFeaturesAreIndexed (entityID) == 0) {
SeqMgrIndexFeatures (entityID, 0);
}
VisitBioseqsInSep (sep, (Pointer) cfp, DoFastaSeq);
}
if (StringChr (cfp->seq, 'r') != NULL) {
VisitBioseqsInSep (sep, (Pointer) cfp, DoFastaRaw);
}
if (StringChr (cfp->seq, 'd') != NULL) {
VisitBioseqsInSep (sep, (Pointer) cfp, DoFastaDefline);
}
if (StringChr (cfp->seq, 'D') != NULL) {
if (SeqMgrFeaturesAreIndexed (entityID) == 0) {
SeqMgrIndexFeatures (entityID, 0);
}
VisitBioseqsInSep (sep, (Pointer) cfp, DoFastaDefline);
}
if (StringChr (cfp->seq, 'T') != NULL) {
VisitDescriptorsInSep (sep, NULL, MarkTitles);
DeleteMarkedObjects (entityID, 0, NULL);
SeqMgrIndexFeatures (entityID, 0);
VisitBioseqsInSep (sep, (Pointer) cfp, DoFastaDefline);
}
if (StringChr (cfp->seq, 'x') != NULL) {
VisitBioseqsInSep (sep, (Pointer) cfp, DoNewFastaDefline);
}
if (StringChr (cfp->seq, 'X') != NULL) {
VisitDescriptorsInSep (sep, NULL, MarkTitles);
DeleteMarkedObjects (entityID, 0, NULL);
SeqMgrIndexFeatures (entityID, 0);
VisitBioseqsInSep (sep, (Pointer) cfp, DoNewFastaDefline);
}
if (StringChr (cfp->seq, 'f') != NULL) {
VisitFeaturesInSep (sep, (Pointer) cfp, DoFastaFeat);
}
if (StringChr (cfp->seq, 't') != NULL) {
VisitFeaturesInSep (sep, (Pointer) cfp, DoFastaTrans);
}
if (StringChr (cfp->feat, 'v') != NULL) {
VisitFeaturesInSep (sep, NULL, DoVisitFeaturesTest);
}
if (StringChr (cfp->feat, 'g') != NULL) {
if (SeqMgrFeaturesAreIndexed (entityID) == 0) {
SeqMgrIndexFeatures (entityID, 0);
}
VisitFeaturesInSep (sep, (Pointer) cfp, DoGeneOverlapPrintTest);
}
if (StringChr (cfp->feat, 'h') != NULL) {
if (SeqMgrFeaturesAreIndexed (entityID) == 0) {
SeqMgrIndexFeatures (entityID, 0);
}
VisitFeaturesInSep (sep, (Pointer) cfp, DoGeneOverlapSpeedTest);
}
if (StringChr (cfp->feat, 'x') != NULL) {
}
if (StringChr (cfp->feat, 'o') != NULL) {
}
if (StringChr (cfp->feat, 'd') != NULL) {
}
if (StringChr (cfp->feat, 't') != NULL) {
SeqEntryToGnbk (sep, NULL, FTABLE_FMT, SEQUIN_MODE, NORMAL_STYLE,
0, 0, SHOW_PROT_FTABLE, NULL, cfp->ofp);
}
if (StringChr (cfp->feat, 's') != NULL) {
if (SeqMgrFeaturesAreIndexed (entityID) == 0) {
SeqMgrIndexFeatures (entityID, 0);
}
cfp->nucbsp = FindNucBioseq (sep);
if (cfp->nucbsp != NULL) {
BioseqToGeneticCode (cfp->nucbsp, &(cfp->genCode), NULL, NULL, NULL, 0, NULL);
SeqIdWrite (cfp->nucbsp->id, id, PRINTID_FASTA_LONG, sizeof (id) - 1);
fprintf (cfp->ofp, "%s\n", id);
VisitBioseqsInSep (sep, (Pointer) cfp, DoSuggestIntervals);
cfp->nucbsp = NULL;
cfp->genCode = 0;
}
}
if (StringChr (cfp->feat, 'S') != NULL) {
if (SeqMgrFeaturesAreIndexed (entityID) == 0) {
SeqMgrIndexFeatures (entityID, 0);
}
cfp->nucbsp = FindNucBioseq (sep);
if (cfp->nucbsp != NULL) {
BioseqToGeneticCode (cfp->nucbsp, &(cfp->genCode), NULL, NULL, NULL, 0, NULL);
SetBatchSuggestNucleotide (cfp->nucbsp, cfp->genCode);
SeqIdWrite (cfp->nucbsp->id, id, PRINTID_FASTA_LONG, sizeof (id) - 1);
fprintf (cfp->ofp, "%s\n", id);
VisitBioseqsInSep (sep, (Pointer) cfp, DoSuggestIntervals);
ClearBatchSuggestNucleotide ();
cfp->nucbsp = NULL;
cfp->genCode = 0;
}
}
if (StringChr (cfp->feat, 'c') != NULL) {
VisitFeaturesInSep (sep, (Pointer) cfp, DoVisitCodingRegions);
}
if (StringChr (cfp->desc, 'b') != NULL) {
}
if (StringChr (cfp->desc, 't') != NULL) {
}
if (StringChr (cfp->verify, 'v') != NULL) {
if (SeqMgrFeaturesAreIndexed (entityID) == 0) {
SeqMgrIndexFeatures (entityID, 0);
}
vsp = ValidStructNew ();
if (vsp != NULL) {
vsp->useSeqMgrIndexes = TRUE;
vsp->suppressContext = TRUE;
vsp->seqSubmitParent = TRUE;
vsp->testLatLonSubregion = TRUE;
oldErrSev = ErrSetMessageLevel (SEV_NONE);
vsp->errfunc = ValidCallback;
vsp->userdata = (Pointer) cfp->ofp;
/* vsp->convertGiToAccn = FALSE; */
ValidateSeqEntry (sep, vsp);
ValidStructFree (vsp);
ErrSetMessageLevel (oldErrSev);
}
}
if (StringChr (cfp->verify, 'b') != NULL) {
if (SeqMgrFeaturesAreIndexed (entityID) == 0) {
SeqMgrIndexFeatures (entityID, 0);
}
SeqEntryToGnbk (sep, NULL, GENBANK_FMT, SEQUIN_MODE, NORMAL_STYLE,
0, 0, 0, NULL, cfp->ofp);
}
if (cfp->ofp != NULL) {
fflush (cfp->ofp);
}
}
static void ProcessMultipleRecord (
CharPtr filename,
CSpeedFlagPtr cfp
)
{
AsnIoPtr aip;
AsnTypePtr atp;
BioseqPtr bsp;
Char buf [41];
Uint2 entityID;
FILE *fp;
SeqEntryPtr fsep;
Char longest [41];
Int4 numrecords, x;
SeqEntryPtr sep;
time_t starttime, stoptime, worsttime;
#ifdef OS_UNIX
Char cmmd [256];
CharPtr gzcatprog;
int ret;
Boolean usedPopen = FALSE;
#endif
if (cfp == NULL) return;
if (StringHasNoText (filename)) return;
#ifndef OS_UNIX
if (cfp->compressed) {
Message (MSG_POSTERR, "Can only decompress on-the-fly on UNIX machines");
return;
}
#endif
#ifdef OS_UNIX
if (cfp->compressed) {
gzcatprog = getenv ("NCBI_UNCOMPRESS_BINARY");
if (gzcatprog != NULL) {
sprintf (cmmd, "%s %s", gzcatprog, filename);
} else {
ret = system ("gzcat -h >/dev/null 2>&1");
if (ret == 0) {
sprintf (cmmd, "gzcat %s", filename);
} else if (ret == -1) {
Message (MSG_POSTERR, "Unable to fork or exec gzcat in ScanBioseqSetRelease");
return;
} else {
ret = system ("zcat -h >/dev/null 2>&1");
if (ret == 0) {
sprintf (cmmd, "zcat %s", filename);
} else if (ret == -1) {
Message (MSG_POSTERR, "Unable to fork or exec zcat in ScanBioseqSetRelease");
return;
} else {
Message (MSG_POSTERR, "Unable to find zcat or gzcat in ScanBioseqSetRelease - please edit your PATH environment variable");
return;
}
}
}
fp = popen (cmmd, /* cfp->binary? "rb" : */ "r");
usedPopen = TRUE;
} else {
fp = FileOpen (filename, cfp->binary? "rb" : "r");
}
#else
fp = FileOpen (filename, cfp->binary? "rb" : "r");
#endif
if (fp == NULL) {
Message (MSG_POSTERR, "FileOpen failed for input file '%s'", filename);
return;
}
aip = AsnIoNew (cfp->binary? ASNIO_BIN_IN : ASNIO_TEXT_IN, fp, NULL, NULL, NULL);
if (aip == NULL) {
Message (MSG_ERROR, "AsnIoNew failed for input file '%s'", filename);
return;
}
if (cfp->logfp != NULL) {
fprintf (cfp->logfp, "%s\n\n", filename);
fflush (cfp->logfp);
}
longest [0] = '\0';
worsttime = 0;
numrecords = 0;
atp = cfp->atp_bss;
while ((atp = AsnReadId (aip, cfp->amp, atp)) != NULL) {
if (atp == cfp->atp_se) {
sep = SeqEntryAsnRead (aip, atp);
if (sep != NULL) {
entityID = ObjMgrGetEntityIDForChoice (sep);
fsep = FindNthBioseq (sep, 1);
if (fsep != NULL && fsep->choice == 1) {
bsp = (BioseqPtr) fsep->data.ptrvalue;
if (bsp != NULL) {
SeqIdWrite (bsp->id, buf, PRINTID_FASTA_LONG, sizeof (buf));
if (cfp->logfp != NULL) {
fprintf (cfp->logfp, "%s\n", buf);
fflush (cfp->logfp);
}
}
}
starttime = GetSecs ();
for (x = 0; x < cfp->maxcount; x++) {
DoProcess (sep, entityID, cfp);
}
stoptime = GetSecs ();
if (stoptime - starttime > worsttime) {
worsttime = stoptime - starttime;
StringCpy (longest, buf);
}
numrecords++;
ObjMgrFreeByEntityID (entityID);
}
} else {
AsnReadVal (aip, atp, NULL);
}
}
AsnIoFree (aip, FALSE);
#ifdef OS_UNIX
if (usedPopen) {
pclose (fp);
} else {
FileClose (fp);
}
#else
FileClose (fp);
#endif
if (cfp->logfp != NULL && (! StringHasNoText (longest))) {
fprintf (cfp->logfp, "Longest processing time %ld seconds on %s\n",
(long) worsttime, longest);
fprintf (cfp->logfp, "Total number of records %ld\n", (long) numrecords);
fflush (cfp->logfp);
}
}
static void ProcessAccession (
CharPtr accn,
XtraPtr extra,
Boolean only_new,
Boolean get_var,
Boolean do_nuc,
Boolean do_prot
)
{
Char ch;
Int4 flags = 0;
Int4 gi = 0;
Char id [41];
Boolean is_numeric = TRUE;
Int4 newgi = 0;
CharPtr ptr;
SeqEntryPtr sep;
SeqIdPtr sip;
Char tmp [41];
long val;
ptr = accn;
ch = *ptr;
while (ch != '\0' && is_numeric) {
if (! IS_DIGIT (ch)) {
is_numeric = FALSE;
}
ptr++;
ch = *ptr;
}
if (is_numeric) {
if (sscanf (accn, "%ld", &val) == 1) {
gi = (Int4) val;
if (gi < 1) return;
if (only_new) {
sip = GetSeqIdForGI (gi);
if (sip != NULL) {
SeqIdWrite (sip, tmp, PRINTID_TEXTID_ACC_VER, sizeof (tmp));
SeqIdFree (sip);
ptr = StringChr (tmp, '.');
if (ptr != NULL) {
*ptr = '\0';
sip = SeqIdFromAccessionDotVersion (tmp);
newgi = GetGIForSeqId (sip);
SeqIdFree (sip);
if (newgi == gi) return;
}
}
}
}
} else {
sip = SeqIdFromAccessionDotVersion (accn);
gi = GetGIForSeqId (sip);
SeqIdFree (sip);
if (only_new) {
sip = GetSeqIdForGI (gi);
if (sip != NULL) {
SeqIdWrite (sip, id, PRINTID_TEXTID_ACC_VER, sizeof (id));
SeqIdFree (sip);
if (StringICmp (accn, id) == 0) return;
}
}
}
if (gi < 1) return;
if (get_var) {
flags = 1;
}
sep = PubSeqSynchronousQuery (gi, 0, flags);
if (sep == NULL) return;
if (do_nuc) {
DoSeqEntryToGnbk (sep, GENBANK_FMT, extra);
}
if (do_prot) {
DoSeqEntryToGnbk (sep, GENPEPT_FMT, extra);
}
SeqEntryFree (sep);
}
/*********************************************************************
*
* make_cds_paragraph(sfp, aa_start, aa_stop)
* return a buffer for the display of 3-codon under one amino
* acid format. It also includes the new line characters
* This is what Jonathan K. desires to have for the sequin
* doc object
* aa_start, aa_stop: start and stop in the amino acid sequence
*
*********************************************************************/
NLM_EXTERN CharPtr make_cds_paragraph(SeqFeatPtr sfp, Int4 aa_start, Int4 aa_stop)
{
BioseqPtr pbsp;
SeqPortPtr spp;
ValNodePtr cvp_node, curr;
CodonVectorPtr cvp;
CharPtr docbuf = NULL;
Int4 num, buf_size;
Uint1 residue;
Char p_name[30];
Int4 space_len, i;
CharPtr buf;
Int4 pos;
Int4 max_len = 150;
Boolean extra_space;
if(sfp == NULL || sfp->data.choice !=3)
return NULL;
if(sfp->product == NULL)
return NULL;
pbsp = BioseqLockById(SeqLocId(sfp->product));
if(pbsp == NULL)
return NULL;
cvp_node = aa_to_codon(sfp, aa_start, aa_stop);
num = 1;
for(curr = cvp_node; curr !=NULL; curr = curr->next)
num +=3;
buf_size = num * max_len;
/* #ifdef WIN_16
if(buf_size > 10000)
{
Message(MSG_ERROR, "Can not allocate enough space ");
return NULL;
}
#endif
*/
docbuf = MemNew((size_t)(buf_size) * sizeof(Char));
MuskSeqIdWrite(pbsp->id, p_name, B_SPACE, PRINTID_TEXTID_ACCESSION, TRUE, FALSE);
/*SeqIdWrite (pbsp->id, p_name, PRINTID_FASTA_SHORT, 10);*/
pos = 0;
pos+= print_label_to_buffer(docbuf+pos, p_name, (aa_start+1), 0, FALSE,
FALSE, B_SPACE, POS_SPACE);
/*print the amino acid sequence into buffer*/
spp = SeqPortNew(pbsp, aa_start, aa_stop, Seq_strand_plus, Seq_code_ncbieaa);
while((residue = SeqPortGetResidue(spp)) != SEQPORT_EOF )
docbuf[pos++] = residue;
docbuf[pos++] = '\n';
SeqPortFree(spp);
for(curr = cvp_node; curr !=NULL; curr = curr->next)
{
cvp = curr->data.ptrvalue;
SeqIdWrite (cvp->sip, p_name, PRINTID_FASTA_SHORT, 10);
extra_space = (cvp->aa_index == 0);
for(i=0; i<3; ++i)
{
space_len = cvp->aa_index;
buf = cvp->buf[i] + cvp->aa_index;
if(i == cvp->frame)
{
pos+= print_label_to_buffer(docbuf+pos, p_name,
cvp->dna_pos, cvp->strand, extra_space, FALSE, B_SPACE, POS_SPACE);
}
else
pos+= print_label_to_buffer(docbuf+pos, NULL, -1,
0, extra_space, FALSE, B_SPACE, POS_SPACE);
sprintf(docbuf+pos, "%s\n", buf);
pos += (StringLen(buf) +1);
}
}
docbuf[pos++] = '\n';
docbuf[pos] = '\0';
free_cvp_list(cvp_node);
BioseqUnlock(pbsp);
return docbuf;
}
/** Creates the header part of an XML report for a BLAST search.
* @param program Program name [in]
* @param database Database name [in]
* @param query_loc Query Seq-loc [in]
* @param flags Flag to indicate whether query sequence should be included in
* the output. [in]
* @param search_param Search parameters [in]
*/
static BlastOutput*
s_CreateBlastOutputHead(const char* program, const char* database,
SeqLoc* query_loc, Int4 flags,
const Blast_SearchParams* search_param)
{
BlastOutput* boutp;
Char buffer[1024];
char* program_to_use = NULL;
if((boutp = BlastOutputNew()) == NULL)
return FALSE;
if (strcmp(program, "rpsblast") == 0)
program_to_use = strdup("blastp");
else if (strcmp(program, "rpstblastn") == 0)
program_to_use = strdup("blastx");
else
program_to_use = strdup(program);
/* For optimization BLOSUM62 may be loaded ones */
if (query_loc) {
SeqId* sip = SeqLocId(query_loc);
Bioseq* bsp;
SeqIdWrite(sip, buffer, PRINTID_FASTA_LONG, sizeof(buffer));
boutp->query_ID = strdup(buffer);
bsp = BioseqLockById(sip);
if(bsp != NULL) {
if (BioseqGetTitle(bsp) != NULL)
boutp->query_def = strdup(BioseqGetTitle(bsp));
else
boutp->query_def = strdup("No definition line found");
}
BioseqUnlock(bsp);
boutp->query_len = SeqLocLen(query_loc);
if(flags & BXML_INCLUDE_QUERY) {
boutp->query_seq = (char *) calloc(boutp->query_len+1, 1);
SeqPortStreamLoc(query_loc,
STREAM_EXPAND_GAPS | STREAM_CORRECT_INVAL,
boutp->query_seq, NULL);
} else {
boutp->query_seq = NULL; /* Do we need sequence here??? */
}
}
/* Program name. Use the local version of the program. No need to copy it
since it was locally allocated. */
boutp->program = program_to_use;
/* Database name */
if (database)
boutp->db = strdup(database);
/* Version text */
sprintf(buffer, "%s %s [%s]", program_to_use, BlastGetVersionNumber(),
BlastGetReleaseDate());
boutp->version = strdup(buffer);
/* Reference */
boutp->reference = BlastGetReference(FALSE);
/* Filling parameters */
boutp->param = ParametersNew();
boutp->param->expect = search_param->expect;
boutp->param->gap_open = search_param->gap_open;
boutp->param->gap_extend = search_param->gap_extension;
if (search_param->matrix)
boutp->param->matrix = strdup(search_param->matrix);
boutp->param->sc_match = search_param->match;
boutp->param->sc_mismatch = search_param->mismatch;
boutp->param->include = search_param->ethresh;
if (search_param->filter_string)
boutp->param->filter = strdup(search_param->filter_string);
return boutp;
}
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;
}
