From 2327244873a9cc88120433bd1c68ea3757d839ea Mon Sep 17 00:00:00 2001
From: gb10 <gb10>
Date: Tue, 5 Oct 2010 15:51:21 +0000
Subject: [PATCH] Added a type for BlxSequence
---
blxFetch.c | 14 +++++-----
blxparser.c | 6 ++---
blxview.c | 78 ++++++++++++++++++++++++++++++++---------------------
blxview.h | 20 +++++++++++---
utilities.c | 18 ++++++++++---
utilities.h | 2 ++
6 files changed, 91 insertions(+), 47 deletions(-)
diff --git a/blxFetch.c b/blxFetch.c
index be818c57..78a784f7 100644
--- a/blxFetch.c
+++ b/blxFetch.c
@@ -38,7 +38,7 @@
* HISTORY:
* Last edited: Aug 21 17:34 2009 (edgrif)
* Created: Tue Jun 17 16:20:26 2008 (edgrif)
- * CVS info: $Id: blxFetch.c,v 1.39 2010-10-05 15:23:02 gb10 Exp $
+ * CVS info: $Id: blxFetch.c,v 1.40 2010-10-05 15:51:21 gb10 Exp $
*-------------------------------------------------------------------
*/
@@ -928,8 +928,8 @@ gboolean populateFastaDataPfetch(GList *seqsToFetch, const char *pfetchIP, int p
* - can be called after the fasta sequence data is already populated:
* in that case it will ignore the sequence data and just populate
* the additional data.
- * - sequence data will also be ignored for sequences that have the
- * sequenceReqd flag set to false
+ * - sequence data will also be ignored for sequences that do not
+ * require sequence data
*/
gboolean populateFullDataPfetch(GList *seqsToFetch, const char *pfetchIP, int port, gboolean External, const BlxSeqType seqType, GError **error)
{
@@ -2155,7 +2155,7 @@ static gboolean pfetchGetParserStateFromId(const char *sectionId,
if (stringsEqual(sectionId, "SQ", TRUE))
{
/* Skip the sequence section if the sequence is already populated or not required. */
- if (currentSeq->sequence && (currentSeq->sequence->str || !currentSeq->sequenceReqd))
+ if (currentSeq->sequence && (currentSeq->sequence->str || !blxSequenceRequiresSeqData(currentSeq)))
{
*parserState = PARSING_IGNORE;
}
@@ -2164,15 +2164,15 @@ static gboolean pfetchGetParserStateFromId(const char *sectionId,
*parserState = PARSING_SEQUENCE_HEADER;
}
}
- else if (stringsEqual(sectionId, "OS", TRUE) && currentSeq->optionalDataReqd)
+ else if (stringsEqual(sectionId, "OS", TRUE) && blxSequenceRequiresOptionalData(currentSeq))
{
*parserState = PARSING_ORGANISM;
}
- else if (stringsEqual(sectionId, "GN", TRUE) && currentSeq->optionalDataReqd)
+ else if (stringsEqual(sectionId, "GN", TRUE) && blxSequenceRequiresOptionalData(currentSeq))
{
*parserState = PARSING_GENE_NAME;
}
- else if (stringsEqual(sectionId, "FT", TRUE) && currentSeq->optionalDataReqd)
+ else if (stringsEqual(sectionId, "FT", TRUE) && blxSequenceRequiresOptionalData(currentSeq))
{
if (tagName && tagName->len)
{
diff --git a/blxparser.c b/blxparser.c
index 63700ad9..c02b0593 100644
--- a/blxparser.c
+++ b/blxparser.c
@@ -34,7 +34,7 @@
* * 98-02-19 Changed MSP parsing to handle all SFS formats.
* * 99-07-29 Added support for SFS type=HSP and GFF.
* Created: 93-05-17
- * CVS info: $Id: blxparser.c,v 1.38 2010-08-27 12:25:14 gb10 Exp $
+ * CVS info: $Id: blxparser.c,v 1.39 2010-10-05 15:51:21 gb10 Exp $
*-------------------------------------------------------------------
*/
@@ -903,7 +903,7 @@ static void parseEXBLXSEQBLExtended(MSP **lastMsp, MSP **mspList, BlxParserState
g_error("Bad description data\n");
}
}
- else if ((parserState == SEQBL_X_BODY || mspIsSnp(msp)) && (strstr(seq_pos, BLX_SEQUENCE_TAG)))
+ else if ((parserState == SEQBL_X_BODY || mspIsVariation(msp)) && (strstr(seq_pos, BLX_SEQUENCE_TAG)))
{
if (*opts == 'L')
{
@@ -1782,7 +1782,7 @@ static BlxMspType getMspTypeFromScore(const int score)
}
else if (score == -3)
{
- result = BLXMSP_SNP;
+ result = BLXMSP_VARIATION;
}
return result;
diff --git a/blxview.c b/blxview.c
index 9799e137..86fce988 100644
--- a/blxview.c
+++ b/blxview.c
@@ -88,7 +88,7 @@
01-10-05 Added getsseqsPfetch to fetch all missing sseqs in one go via socket connection to pfetch [RD]
* Created: Thu Feb 20 10:27:39 1993 (esr)
- * CVS info: $Id: blxview.c,v 1.73 2010-10-05 15:23:02 gb10 Exp $
+ * CVS info: $Id: blxview.c,v 1.74 2010-10-05 15:51:21 gb10 Exp $
*-------------------------------------------------------------------
*/
@@ -811,6 +811,28 @@ static gint compareFuncMspPos(gconstpointer a, gconstpointer b)
return result;
}
+
+static BlxSequenceType getBlxSequenceTypeForMsp(const MSP const *msp)
+{
+ BlxSequenceType result = BLXSEQUENCE_UNSET;
+
+ if (mspIsBlastMatch(msp))
+ {
+ result = BLXSEQUENCE_MATCH;
+ }
+ else if (mspIsExon(msp) || mspIsIntron(msp))
+ {
+ result = BLXSEQUENCE_TRANSCRIPT;
+ }
+ else if (mspIsVariation(msp))
+ {
+ result = BLXSEQUENCE_VARIATION;
+ }
+
+ return result;
+}
+
+
/* Add or create a BlxSequence struct, creating the BlxSequence if one does not
* already exist for the MSP's sequence name. Seperate BlxSequence structs are created
* for the forward and reverse strands of the same sequence. The passed-in sequence
@@ -839,12 +861,7 @@ BlxSequence* addBlxSequence(const char *name, const char *idTag, BlxStrand stran
/* Create a new BlxSequence, and take ownership of the passed in sequence (if any) */
blxSeq = createEmptyBlxSequence(seqName, idTag, NULL);
*seqList = g_list_prepend(*seqList, blxSeq);
-
blxSeq->strand = strand;
-
- /* Set whether the sequence data is required by any of this sequence's MSPs */
- blxSeq->sequenceReqd |= mspIsBlastMatch(msp) || mspIsVariation(msp);
- blxSeq->optionalDataReqd |= mspIsBlastMatch(msp);
}
if (seqName && !blxSeq->fullName)
@@ -859,6 +876,15 @@ BlxSequence* addBlxSequence(const char *name, const char *idTag, BlxStrand stran
/* Add the MSP to the BlxSequence's list. Keep it sorted by position. */
blxSeq->mspList = g_list_insert_sorted(blxSeq->mspList, msp, compareFuncMspPos);
msp->sSequence = blxSeq;
+
+ if (blxSeq->type == BLXSEQUENCE_UNSET)
+ {
+ blxSeq->type = getBlxSequenceTypeForMsp(msp);
+ }
+ else if (blxSeq->type != getBlxSequenceTypeForMsp(msp))
+ {
+ g_warning("Adding MSP of type %d to parent of type %d (expected parent type to be %d)\n", msp->type, blxSeq->type, getBlxSequenceTypeForMsp(msp));
+ }
}
/* Add the sequence data */
@@ -1366,7 +1392,7 @@ void addBlxSequenceData(BlxSequence *blxSeq, char *sequence, GError **error)
gboolean sequenceUsed = FALSE;
- if (blxSeq && blxSeq->sequenceReqd)
+ if (blxSeq && blxSequenceRequiresSeqData(blxSeq))
{
if (!blxSeq->sequence)
{
@@ -1644,31 +1670,23 @@ GList* getSeqsToPopulate(GList *inputList, const gboolean getSequenceData, const
{
BlxSequence *blxSeq = (BlxSequence*)(inputItem->data);
- gboolean getSeq = FALSE;
+ /* Check if sequence data was requested and is not already set. */
+ gboolean getSeq = (blxSequenceRequiresSeqData(blxSeq) && getSequenceData && blxSeq->sequence == NULL);
+
+ /* Check if optional data was requested and is not already set. We can assume that
+ * if any of the data fields is set then the parsing has been done for all of them
+ * (and any remaining empty fields just don't have that data available) */
+ getSeq |= (blxSequenceRequiresOptionalData(blxSeq) &&
+ getOptionalData &&
+ !blxSeq->organism &&
+ !blxSeq->geneName &&
+ !blxSeq->tissueType &&
+ !blxSeq->strain);
- /* We only want to get data for blast matches, which have the sequenceReqd flag set. */
- if (blxSeq->sequenceReqd)
+ if (getSeq)
{
- /* Check if sequence data was requested and is not already set. */
- getSeq = (getSequenceData && blxSeq->sequence == NULL);
-
- if (blxSeq->optionalDataReqd)
- {
- /* Check if optional data was requested and is not already set. We can assume that
- * if any of the data fields is set then the parsing has been done for all of them
- * (and any remaining empty fields just don't have that data available) */
- getSeq |= (getOptionalData &&
- !blxSeq->organism &&
- !blxSeq->geneName &&
- !blxSeq->tissueType &&
- !blxSeq->strain);
- }
-
- if (getSeq)
- {
- resultList = g_list_prepend(resultList, blxSeq);
- }
- }
+ resultList = g_list_prepend(resultList, blxSeq);
+ }
}
return resultList;
diff --git a/blxview.h b/blxview.h
index 1b0029e1..97b7ae45 100644
--- a/blxview.h
+++ b/blxview.h
@@ -27,7 +27,7 @@
* Last edited: Aug 21 13:57 2009 (edgrif)
* * Aug 26 16:57 1999 (fw): added this header
* Created: Thu Aug 26 16:57:17 1999 (fw)
- * CVS info: $Id: blxview.h,v 1.37 2010-10-05 15:23:02 gb10 Exp $
+ * CVS info: $Id: blxview.h,v 1.38 2010-10-05 15:51:21 gb10 Exp $
*-------------------------------------------------------------------
*/
#ifndef DEF_BLXVIEW_H
@@ -86,9 +86,23 @@ typedef enum
} BlxStrand ;
-/* Structure that contains information about a sequence */
+/* Type definition for BlxSequences */
+typedef enum
+{
+ BLXSEQUENCE_UNSET,
+ BLXSEQUENCE_TRANSCRIPT, /* transcript (i.e. collection of exons and introns) */
+ BLXSEQUENCE_MATCH, /* match sequence (i.e. collection of matches) */
+ BLXSEQUENCE_VARIATION /* variation (i.e. insertion, deletion or substitution) */
+} BlxSequenceType;
+
+
+/* Structure that contains information about a "sequence" (bit of a misnomer because it also includes
+ * transcripts etc. for which it doesn't actually contain any sequence data. Really this is just a
+ * parent object for collections of MSPs). */
typedef struct _BlxSequence
{
+ BlxSequenceType type; /* What type of collection of MSPs this is */
+
char *idTag; /* Unique identifier e.g. from ID tag in GFF files */
char *fullName; /* full name of the sequence and variant, including prefix characters, e.g. EM:AV274505.2 */
@@ -102,8 +116,6 @@ typedef struct _BlxSequence
BlxStrand strand; /* which strand of the sequence this is */
GString *sequence; /* the actual sequence data */
- gboolean sequenceReqd; /* whether the sequence data is required (e.g. it is not needed for exons/introns etc.) */
- gboolean optionalDataReqd; /* whether the optional data is required (e.g. it is not applicable to exons/snps etc.) */
IntRange qRange; /* the extent of the sequence on the ref sequence */
GList *mspList; /* list of MSPs from this sequence */
diff --git a/utilities.c b/utilities.c
index 80f09fd2..ddc575ea 100644
--- a/utilities.c
+++ b/utilities.c
@@ -1794,7 +1794,8 @@ BlxSequence* createEmptyBlxSequence(const char *fullName, const char *idTag, GEr
}
BlxSequence *seq = g_malloc(sizeof(BlxSequence));
-
+
+ seq->type = BLXSEQUENCE_UNSET;
seq->idTag = idTag ? g_strdup(idTag) : NULL;
seq->fullName = NULL;
@@ -1804,8 +1805,6 @@ BlxSequence* createEmptyBlxSequence(const char *fullName, const char *idTag, GEr
seq->mspList = NULL;
seq->sequence = NULL;
- seq->sequenceReqd = FALSE;
- seq->optionalDataReqd = FALSE;
seq->organism = NULL;
seq->geneName = NULL;
@@ -1816,6 +1815,19 @@ BlxSequence* createEmptyBlxSequence(const char *fullName, const char *idTag, GEr
}
+/* Returns true if the given BlxSequence requires sequence data to be set. */
+gboolean blxSequenceRequiresSeqData(const BlxSequence *blxSeq)
+{
+ return (blxSeq->type == BLXSEQUENCE_MATCH || blxSeq->type == BLXSEQUENCE_VARIATION);
+}
+
+/* Returns true if the given BlxSequence uses optional data. */
+gboolean blxSequenceRequiresOptionalData(const BlxSequence *blxSeq)
+{
+ return (blxSeq->type == BLXSEQUENCE_MATCH);
+}
+
+
/* Round to nearest int (needed because there is no round() function in ISO C90) */
int roundNearest(const double val)
{
diff --git a/utilities.h b/utilities.h
index 04d25cac..41d28f1c 100644
--- a/utilities.h
+++ b/utilities.h
@@ -221,6 +221,8 @@ BlxStrand mspGetMatchStrand(const MSP const *msp);
const char* mspGetMatchSeq(const MSP const *msp);
const char* mspGetSName(const MSP *msp);
char* blxSequenceGetSummaryInfo(const BlxSequence const *blxSeq);
+gboolean blxSequenceRequiresSeqData(const BlxSequence *blxSeq);
+gboolean blxSequenceRequiresOptionalData(const BlxSequence *blxSeq);
char* mspGetExonTranscriptName(const MSP *msp);
const IntRange const* mspGetRefCoords(const MSP const *msp);
const IntRange const* mspGetMatchCoords(const MSP const *msp);
--
GitLab