0513e85cc01843ecc411b86dfce9180a8d1acf37
max
Thu Mar 22 17:40:16 2012 -0700
make individual sequences for publications track clickable, show alignments, correct mouseOvers // Also changs text that is shown for PDF Output in hgTracks (see email with Ann/Donna/Brian)
diff --git src/hg/hgc/pubs.c src/hg/hgc/pubs.c
index cd09f14..41e8f4a 100644
--- src/hg/hgc/pubs.c
+++ src/hg/hgc/pubs.c
@@ -1,49 +1,63 @@
/* pubs.c - display details of publiations literature track (pubsxxx tables) */
#include "common.h"
#include "jksql.h"
#include "hdb.h"
#include "hgc.h"
#include "hgColors.h"
#include "trackDb.h"
#include "web.h"
#include "hash.h"
#include "obscure.h"
+#include "common.h"
+#include "string.h"
//include "hgTrackUi.h"
// cgi var to activate debug output
static int pubsDebug = 0;
// internal section types in mysql table
static char* pubsSecNames[] ={
"header", "abstract",
"intro", "methods",
"results", "discussion",
"conclusions", "ack",
"refs", "unknown" };
//
// whether a checkbox is checked by default, have to correspond to pubsSecNames
static int pubsSecChecked[] ={
1, 1,
1, 1,
1, 1,
1, 0,
0, 1 };
static char* pubsSequenceTable;
-static char* pubsArticleTable;
+
+static void printFilterLink(char* pslTrack, char* articleId)
+/* print a link to hgTracks with an additional cgi param to activate the single article filter */
+{
+ int start = cgiInt("o");
+ int end = cgiInt("t");
+ printf(" <A HREF=\"%s&db=%s&position=%s%%3A%d-%d&pubsFilterArticleId=%s&%s=pack\">",
+ hgTracksPathAndSettings(), database, seqName, start+1, end, articleId, pslTrack);
+ char startBuf[64], endBuf[64];
+ sprintLongWithCommas(startBuf, start + 1);
+ sprintLongWithCommas(endBuf, end);
+ printf("Show these sequence matches individually on genome browser</A>");
+}
static char* makeSqlMarkerList(void)
/* return list of sections from cgi vars, format like "'abstract','header'" */
{
int secCount = sizeof(pubsSecNames)/sizeof(char *);
struct slName* names = NULL;
int i;
for (i=0; i<secCount; i++)
{
// add ' around name and add to list
char* secName = pubsSecNames[i];
if (cgiOptionalInt(secName, pubsSecChecked[i]))
{
char nameBuf[100];
safef(nameBuf, sizeof(nameBuf), "'%s'", secName);
@@ -173,31 +187,31 @@
printf("<SMALL>%s</SMALL>; ", authors);
printf("<SMALL>%s ", citation);
if (!isEmpty(pmid) && strcmp(pmid, "0")!=0 )
printf(", <A HREF=\"http://www.ncbi.nlm.nih.gov/pubmed/%s\">PMID%s</A>\n", pmid, pmid);
printf("</SMALL><BR>\n");
if (pubsDebug)
printf("articleId=%s", articleId);
printf("<I>%s</I><P>", snippets);
printf("<HR>");
}
freeMem(sectionList);
sqlFreeResult(&sr);
}
-static char* printArticleInfo(struct sqlConnection *conn, char* item)
+static char* printArticleInfo(struct sqlConnection *conn, char* item, char* pubsArticleTable)
/* Header with information about paper, return documentId */
{
char query[512];
safef(query, sizeof(query), "SELECT articleId, url, title, authors, citation, abstract, pmid FROM %s WHERE articleId='%s'", pubsArticleTable, item);
struct sqlResult *sr = sqlGetResult(conn, query);
char **row;
char *articleId=NULL;
if ((row = sqlNextRow(sr)) == NULL)
{
printf("Could not resolve articleId %s, this is an internal error.\n", item);
printf("Please send an email to max@soe.ucsc.edu\n");
sqlFreeResult(&sr);
return NULL;
@@ -238,211 +252,257 @@
return NULL;
}
/* get sequence-Ids for feature that was clicked (item&startPos are unique) and return as hash*/
safef(query, sizeof(query), "SELECT seqIds,'' FROM %s WHERE name='%s' "
"and chrom='%s' and chromStart=%d;", trackTable, item, seqName, start);
if (pubsDebug)
printf("%s<br>", query);
// split comma-sep list into parts
char* seqIdCoordString = sqlQuickString(conn, query);
char* seqIdCoords[1024];
int partCount = chopString(seqIdCoordString, ",", seqIdCoords, ArraySize(seqIdCoords));
int i;
- // format of part <fileId:3><annotId:5>:<start>-<end>, e.g. 12312345:0-23
struct hash *seqIdHash = NULL;
seqIdHash = newHash(0);
for (i=0; i<partCount; i++)
{
- char* annotIdHitCoord[1024];
- chopString(seqIdCoords[i], ":", annotIdHitCoord, ArraySize(annotIdHitCoord));
- // put articleId and annotId together to create annotation(=sequence) Id
- char fullAnnotId[4096];
- safef(fullAnnotId, sizeof(fullAnnotId), "%s%s", articleId, annotIdHitCoord[0]);
- if (pubsDebug)
- printf("%s, %s, %s<br>", annotIdHitCoord[0], annotIdHitCoord[1], fullAnnotId);
-
- hashAdd(seqIdHash, fullAnnotId, annotIdHitCoord[1]);
+ hashAdd(seqIdHash, seqIdCoords[i], NULL);
}
- freeMem(seqIdCoordString);
return seqIdHash;
}
static void printSeqHeaders(bool showDesc, bool isClickedSection)
{
printf("<TABLE style=\"background-color: #%s\" WIDTH=\"100%%\" CELLPADDING=\"2\">\n", HG_COL_BORDER);
printf("<TR style=\"background-color: #%s; color: #FFFFFF\">\n", HG_COL_TABLE_LABEL);
if (showDesc)
puts(" <TH style=\"width: 10%\">Article file</TH>\n");
- puts(" <TH style=\"width: 70%\">One table row per sequence, with flanking text, sequence in bold</TH>\n");
+ puts(" <TH style=\"width: 60%\">One row per sequence, with flanking text, sequence in bold</TH>\n");
if (pubsDebug)
puts(" <TH style=\"width: 30%\">Identifiers</TH>\n");
if (!isClickedSection && !pubsDebug)
- puts(" <TH style=\"width: 20%\">Feature that includes this match</TH>\n");
+ puts(" <TH style=\"width: 20%\">Chained matches with this sequence</TH>\n");
puts("</TR>\n");
}
static void printAddWbr(char* text, int distance)
/* a crazy hack for firefox/mozilla that is unable to break long words in tables
* We need to add a <wbr> tag every x characters in the text to make text breakable.
*/
{
int i;
i = 0;
char* c;
c = text;
bool doNotBreak = FALSE;
while (*c != 0) {
{
if ((*c=='&') || (*c=='<'))
doNotBreak = TRUE;
if (*c==';' || (*c =='>'))
doNotBreak = FALSE;
printf("%c", *c);
if (i % distance == 0 && ! doNotBreak)
printf("<wbr>");
c++;
i++;
}
}
}
-static bool printSeqSection(char* articleId, char* title, bool showDesc, struct sqlConnection* conn, struct hash* clickedSeqs, bool isClickedSection, bool fasta)
+void printHgTracksLink(char* db, char* chrom, int start, int end, char* linkText, char* optUrlStr)
+/* print link to hgTracks for db at pos */
+{
+char buf[1024];
+if (linkText==NULL)
+{
+ char startBuf[64], endBuf[64];
+ sprintLongWithCommas(startBuf, start + 1);
+ sprintLongWithCommas(endBuf, end);
+ safef(buf, sizeof(buf), "%s:%s-%s (%s)", seqName, startBuf, endBuf, db);
+ linkText = buf;
+}
+
+if (optUrlStr==NULL)
+ optUrlStr = "";
+
+printf("<A HREF=\"%s&db=%s&position=%d-%d&%s\">%s</A>\n", hgTracksPathAndSettings(), db, start, end, optUrlStr, linkText);
+}
+
+void printGbLinks(struct slName* locs)
+/* print hash keys in format hg19/chr1:1-1000 as links */
+{
+struct slName *el;
+for (el = locs; el != NULL; el = el->next)
+{
+ char* locString = el->name;
+ char* db = cloneNextWordByDelimiter(&locString, '/');
+ char* chrom = cloneNextWordByDelimiter(&locString, ':');
+ char* startStr = cloneNextWordByDelimiter(&locString, '-');
+ char* endStr = locString;
+
+ int start = atoi(startStr);
+ int end = atoi(endStr);
+ printHgTracksLink(db, chrom, start, end, NULL, NULL);
+ printf("<BR>");
+ //freeMem(endStr); //XX why can't I free these?
+ //freeMem(chrom);
+ //freeMem(startStr);
+ //freeMem(db);
+}
+}
+
+static bool printSeqSection(char* articleId, char* title, bool showDesc, struct sqlConnection* conn, struct hash* clickedSeqs, bool isClickedSection, bool fasta, char* pslTable)
/* print a table of sequences, show only sequences with IDs in hash,
* There are two sections, respective sequences are shown depending on isClickedSection and clickedSeqs
* - seqs that were clicked on (isClickedSection=True) -> show only seqs in clickedSeqs
* - other seqs (isClickedSection=False) -> show all other seqs
*
* */
{
// get data from mysql
char query[4096];
safef(query, sizeof(query),
"SELECT fileDesc, snippet, locations, articleId, fileId, seqId, sequence "
"FROM %s WHERE articleId='%s';", pubsSequenceTable, articleId);
if (pubsDebug)
puts(query);
struct sqlResult *sr = sqlGetResult(conn, query);
// construct title for section
+ char* otherFormat = NULL;
+ if (fasta)
+ otherFormat = "table";
+ else
+ otherFormat = "fasta";
+
char fullTitle[5000];
safef(fullTitle, sizeof(fullTitle),
- "%s <A HREF=\"../cgi-bin/hgc?%s&o=%s&t=%s&g=%s&i=%s&fasta=%d\"><SMALL>(switch fasta format)</SMALL></A>",
+ "%s <A HREF=\"../cgi-bin/hgc?%s&o=%s&t=%s&g=%s&i=%s&fasta=%d\"><SMALL>(%s format)</SMALL></A>",
title, cartSidUrlString(cart), cgiString("o"), cgiString("t"), cgiString("g"), cgiString("i"),
- !fasta);
+ !fasta, otherFormat);
webNewSection(fullTitle);
if (!fasta)
printSeqHeaders(showDesc, isClickedSection);
char **row;
bool foundSkippedRows = FALSE;
while ((row = sqlNextRow(sr)) != NULL)
{
char* fileDesc = row[0];
char* snippet = row[1];
- char* locList = row[2];
+ char* locString= row[2];
char* artId = row[3];
char* fileId = row[4];
char* seqId = row[5];
char* seq = row[6];
// annotation (=sequence) ID is a 64 bit int with 10 digits for
// article, 3 digits for file, 5 for annotation
char annotId[100];
safef(annotId, 100, "%010d%03d%05d", atoi(artId), atoi(fileId), atoi(seqId));
if (pubsDebug)
printf("%s", annotId);
// only display this sequence if we're in the right section
if (clickedSeqs!=NULL && ((hashLookup(clickedSeqs, annotId)!=NULL) != isClickedSection)) {
foundSkippedRows = TRUE;
continue;
}
if (fasta)
{
- printf("<TT>>%s<BR>%s<BR></TT>", annotId, seq);
+ printf("<TR><TD><TT>>%s<BR>%s<BR></TT></TD></TR></TABLE>", annotId, seq);
}
else
{
printf("<TR style=\"background-color: #%s\">\n", HG_COL_LOCAL_TABLE);
if (showDesc)
printf("<TD style=\"word-break:break-all\">%s\n", fileDesc);
//printf("<TD><I>%s</I></TD>\n", snippet);
printf("<TD style=\"word-break:break-all;\"><I>");
printAddWbr(snippet, 40);
printf("</I></TD>\n");
if (pubsDebug)
{
printf("<TD>article %s, file %s, seq %s, annotId %s", artId, fileId, seqId, annotId);
}
// print links to locations
if (!isClickedSection && !pubsDebug) {
- struct slName *locs;
// format: hg19/chr1:300-400,mm9/chr1:60006-23234
// split on "," then split on "/"
- locs = charSepToSlNames(locList, ',');
+ //locs = charSepToSlNames(locString, ',');
+
+ char* locArr[1024];
+ int partCount = chopString(locString, ",", locArr, ArraySize(locArr));
printf("<TD>");
- if (locs==NULL)
+ if (partCount==0)
printf("No matches");
- for ( ; locs!=NULL; locs = locs->next)
- {
- char* locString = locs->name;
- char* parts[2];
- int partCount;
- partCount = chopString(locString, "/", parts, ArraySize(parts));
- assert(partCount==2);
- char* db = parts[0];
- char* pos = parts[1];
- printf("<A HREF=\"../cgi-bin/hgTracks?%s&db=%s&position=%s\">%s (%s)</A>", cartSidUrlString(cart), db, pos, pos, db);
+
+ else
+ {
+ struct slName *locs;
+ locs = slNameListFromStringArray(locArr, partCount);
+ slUniqify(&locs, slNameCmp, slNameFree);
+ printGbLinks(locs);
printf("<BR>");
- }
printf("</TD>\n");
+ slFreeList(&locs);
+ }
+
}
printf("</TR>\n");
}
}
printf("</TR>\n");
+
+ if (isClickedSection)
+ {
+ printf("</TABLE></TABLE><TR><TD><P> ");
+ printFilterLink(pslTable, articleId);
+ }
webEndSectionTables();
sqlFreeResult(&sr);
return foundSkippedRows;
}
static void printSeqInfo(struct sqlConnection* conn, char* trackTable,
- char* articleId, char* item, char* seqName, int start, bool fileDesc, bool fasta)
+ char* pslTable, char* articleId, char* item, char* seqName, int start,
+ bool fileDesc, bool fasta)
/* print sequences, split into two sections
* two sections: one for sequences that were clicked, one for all others*/
{
struct hash* clickedSeqs = getSeqIdHash(conn, trackTable, articleId, item, seqName, start);
bool skippedRows;
if (clickedSeqs)
skippedRows = printSeqSection(articleId, "Sequences used to construct this feature", \
- fileDesc, conn, clickedSeqs, 1, fasta);
+ fileDesc, conn, clickedSeqs, 1, fasta, pslTable);
else
skippedRows=1;
if (skippedRows)
printSeqSection(articleId, "Other Sequences in this article", \
- fileDesc, conn, clickedSeqs, 0, fasta);
+ fileDesc, conn, clickedSeqs, 0, fasta, pslTable);
printf("<P><SMALL>Copyright 2012 Elsevier B.V. All rights reserved.</SMALL><P>");
freeHash(&clickedSeqs);
}
static void printTrackVersion(struct trackDb *tdb, struct sqlConnection* conn, char* item)
{
char versionString[256];
char dateReference[256];
char headerTitle[512];
/* see if hgFixed.trackVersion exists */
boolean trackVersionExists = hTableExists("hgFixed", "trackVersion");
if (trackVersionExists)
{
@@ -481,55 +541,116 @@
static void printPositionAndSize(int start, int end, bool showSize)
{
printf("<B>Position:</B> "
"<A HREF=\"%s&db=%s&position=%s%%3A%d-%d\">",
hgTracksPathAndSettings(), database, seqName, start+1, end);
char startBuf[64], endBuf[64];
sprintLongWithCommas(startBuf, start + 1);
sprintLongWithCommas(endBuf, end);
printf("%s:%s-%s</A><BR>\n", seqName, startBuf, endBuf);
long size = end - start;
sprintLongWithCommas(startBuf, size);
if (showSize)
printf("<B>Genomic Size:</B> %s<BR>\n", startBuf);
}
+static bioSeq *getSeq(struct sqlConnection *conn, char *table, char *id)
+/* copied from otherOrgs.c */
+{
+char query[512];
+struct sqlResult *sr;
+char **row;
+bioSeq *seq = NULL;
+safef(query, sizeof(query),
+ "select sequence from %s where annotId = '%s'", table, id);
+sr = sqlGetResult(conn, query);
+if ((row = sqlNextRow(sr)) != NULL)
+ {
+ AllocVar(seq);
+ seq->name = cloneString(id);
+ seq->dna = cloneString(row[0]);
+ seq->size = strlen(seq->dna);
+ }
+sqlFreeResult(&sr);
+return seq;
+}
+
+void pubsAli(struct sqlConnection *conn, char *pslTable, char *seqTable, char *item)
+/* this is just a ripoff from htcCdnaAli, similar to markd's transMapAli */
+{
+bioSeq *oSeq = NULL;
+writeFramesetType();
+puts("<HTML>");
+printf("<HEAD>\n<TITLE>Literature Sequence vs Genomic</TITLE>\n</HEAD>\n\n");
+
+struct psl *psl = getAlignments(conn, pslTable, item);
+if (psl == NULL)
+ errAbort("Couldn't find alignment at %s:%s", pslTable, item);
+
+oSeq = getSeq(conn, seqTable, item);
+if (oSeq == NULL)
+ errAbort("%s is in pslTable but not in sequence table. Internal error.", item);
+showSomeAlignment(psl, oSeq, gftDna, 0, oSeq->size, NULL, 0, 0);
+printf("hihi");
+}
+
void doPubsDetails(struct trackDb *tdb, char *item)
/* publications custom display */
{
int start = cgiInt("o");
-int end = cgiInt("t");
+int end = cgiOptionalInt("t", 0);
char* trackTable = cgiString("g");
+char* aliTable = cgiOptionalString("aliTable");
int fasta = cgiOptionalInt("fasta", 0);
-
pubsDebug = cgiOptionalInt("debug", 0);
struct sqlConnection *conn = hAllocConn(database);
-printTrackVersion(tdb, conn, item);
-if (hashFindVal(tdb->settingsHash, "pubsMarkerTable"))
+char* articleTable = trackDbRequiredSetting(tdb, "pubsArticleTable");
+
+if (stringIn("Psl", trackTable))
+{
+ if (aliTable!=NULL)
+ {
+ pubsSequenceTable = trackDbRequiredSetting(tdb, "pubsSequenceTable");
+ pubsAli(conn, trackTable, pubsSequenceTable, item);
+ return;
+ }
+
+ else
+ {
+ genericHeader(tdb, item);
+ struct psl *psl = getAlignments(conn, trackTable, item);
+ printf("<H3>Genomic Alignment with sequence found in publication fulltext</H3>");
+ printAlignmentsSimple(psl, start, trackTable, trackTable, item);
+ }
+}
+
+else
+{
+ printTrackVersion(tdb, conn, item);
+ if (trackDbSettingClosestToHome(tdb, "pubsMarkerTable") != NULL)
{
char* markerTable = hashMustFindVal(tdb->settingsHash, "pubsMarkerTable");
- char* articleTable = hashMustFindVal(tdb->settingsHash, "pubsArticleTable");
printPositionAndSize(start, end, 0);
printMarkerSnippets(conn, articleTable, markerTable, item);
}
else
{
printPositionAndSize(start, end, 1);
pubsSequenceTable = hashMustFindVal(tdb->settingsHash, "pubsSequenceTable");
- pubsArticleTable = hashMustFindVal(tdb->settingsHash, "pubsArticleTable");
-
- char* articleId = printArticleInfo(conn, item);
+ char* articleId = printArticleInfo(conn, item, articleTable);
if (articleId!=NULL)
{
bool showDesc;
showDesc = (! endsWith(trackTable, "Elsevier"));
// avoid clutter: Elsevier has only main text
- printSeqInfo(conn, trackTable, articleId, item, seqName, start, showDesc, fasta);
+ char *pslTable = trackDbRequiredSetting(tdb, "pubsPslTrack");
+ printSeqInfo(conn, trackTable, pslTable, articleId, item, seqName, start, showDesc, fasta);
+ }
}
}
printTrackHtml(tdb);
hFreeConn(&conn);
}