85f081b107ba60cdb14fb72d2dc439f4e29815df angie Fri Oct 28 14:30:48 2011 -0700 MLQ #5795 (question mark in repeats table): user asked about the ?'s thatappear at the end of some repClass and repFamily names in rmsk. I asked Robert Hubley and he said they are just an indication that a new repeat is suspected to be in some class but hasn't been confirmed yet. He suggested ignoring the ?. I looked at rmskTrack.c and found that the ? at end was causing such items to default to the Other category. This change strips a trailing ? from the class name that we look up in the hash (but leaves it in the ro.repClass field because that's used for the mouseover title). Now if repClass is "DNA?" the item will still be drawn in the DNA row. diff --git src/hg/hgTracks/rmskTrack.c src/hg/hgTracks/rmskTrack.c index 0835b44..661b3c8 100644 --- src/hg/hgTracks/rmskTrack.c +++ src/hg/hgTracks/rmskTrack.c @@ -1,182 +1,187 @@ /* rmskTrack - Handle RepeatMasker track. */ #include "common.h" #include "hash.h" #include "linefile.h" #include "jksql.h" #include "hdb.h" #include "hgTracks.h" #include "rmskOut.h" /* Repeat items. Since there are so many of these, to avoid * memory problems we don't query the database and store the results * during repeatLoad, but rather query the database during the * actual drawing. */ static struct repeatItem *otherRepeatItem = NULL; static char *repeatClassNames[] = { "SINE", "LINE", "LTR", "DNA", "Simple", "Low Complexity", "Satellite", "RNA", "Other", "Unknown", }; static char *repeatClasses[] = { "SINE", "LINE", "LTR", "DNA", "Simple_repeat", "Low_complexity", "Satellite", "RNA", "Other", "Unknown", }; static struct repeatItem *makeRepeatItems() /* Make the stereotypical repeat masker tracks. */ { struct repeatItem *ri, *riList = NULL; int i; int numClasses = ArraySize(repeatClasses); for (i=0; i<numClasses; ++i) { AllocVar(ri); ri->class = repeatClasses[i]; ri->className = repeatClassNames[i]; slAddHead(&riList, ri); if (sameString(repeatClassNames[i], "Other")) otherRepeatItem = ri; } slReverse(&riList); return riList; } static void repeatLoad(struct track *tg) /* Load up repeat tracks. (Will query database during drawing for a change.) */ { tg->items = makeRepeatItems(); } static void repeatFree(struct track *tg) /* Free up repeatMasker items. */ { slFreeList(&tg->items); } static char *repeatName(struct track *tg, void *item) /* Return name of repeat item track. */ { struct repeatItem *ri = item; return ri->className; } static void repeatDraw(struct track *tg, int seqStart, int seqEnd, struct hvGfx *hvg, int xOff, int yOff, int width, MgFont *font, Color color, enum trackVisibility vis) { int baseWidth = seqEnd - seqStart; struct repeatItem *ri; int y = yOff; int heightPer = tg->heightPer; int lineHeight = tg->lineHeight; int x1,x2,w; boolean isFull = (vis == tvFull); Color col; struct sqlConnection *conn = hAllocConn(database); struct sqlResult *sr = NULL; char **row; int rowOffset; if (isFull) { /* Do gray scale representation spread out among tracks. */ struct hash *hash = newHash(6); struct rmskOut ro; int percId; int grayLevel; char statusLine[128]; for (ri = tg->items; ri != NULL; ri = ri->next) { ri->yOffset = y; y += lineHeight; hashAdd(hash, ri->class, ri); } sr = hRangeQuery(conn, tg->table, chromName, winStart, winEnd, NULL, &rowOffset); while ((row = sqlNextRow(sr)) != NULL) { rmskOutStaticLoad(row+rowOffset, &ro); - if (endsWith(ro.repClass, "RNA")) - ri = hashFindVal(hash, "RNA"); - else - ri = hashFindVal(hash, ro.repClass); + char class[256]; + // Simplify repClass for lookup: strip trailing '?', simplify *RNA to RNA: + safecpy(class, sizeof(class), ro.repClass); + char *p = &(class[strlen(class)-1]); + if (*p == '?') + *p = '\0'; + if (endsWith(class, "RNA")) + safecpy(class, sizeof(class), "RNA"); + ri = hashFindVal(hash, class); if (ri == NULL) ri = otherRepeatItem; percId = 1000 - ro.milliDiv - ro.milliDel - ro.milliIns; grayLevel = grayInRange(percId, 500, 1000); col = shadesOfGray[grayLevel]; x1 = roundingScale(ro.genoStart-winStart, width, baseWidth)+xOff; x2 = roundingScale(ro.genoEnd-winStart, width, baseWidth)+xOff; w = x2-x1; if (w <= 0) w = 1; hvGfxBox(hvg, x1, ri->yOffset, w, heightPer, col); if (baseWidth <= 100000) { if (ri == otherRepeatItem) { sprintf(statusLine, "Repeat %s, family %s, class %s", ro.repName, ro.repFamily, ro.repClass); } else { sprintf(statusLine, "Repeat %s, family %s", ro.repName, ro.repFamily); } mapBoxHc(hvg, ro.genoStart, ro.genoEnd, x1, ri->yOffset, w, heightPer, tg->track, ro.repName, statusLine); } } freeHash(&hash); } else { char table[64]; boolean hasBin; struct dyString *query = newDyString(1024); /* Do black and white on single track. Fetch less than we need from database. */ if (hFindSplitTable(database, chromName, tg->table, table, &hasBin)) { dyStringPrintf(query, "select genoStart,genoEnd from %s where ", table); if (hasBin) hAddBinToQuery(winStart, winEnd, query); dyStringPrintf(query, "genoStart<%u and genoEnd>%u ", winEnd, winStart); /* if we're using a single rmsk table, add genoName to the where clause */ if (startsWith("rmsk", table)) dyStringPrintf(query, " and genoName = '%s' ", chromName); sr = sqlGetResult(conn, query->string); while ((row = sqlNextRow(sr)) != NULL) { int start = sqlUnsigned(row[0]); int end = sqlUnsigned(row[1]); x1 = roundingScale(start-winStart, width, baseWidth)+xOff; x2 = roundingScale(end-winStart, width, baseWidth)+xOff; w = x2-x1; if (w <= 0) w = 1; hvGfxBox(hvg, x1, yOff, w, heightPer, MG_BLACK); } } dyStringFree(&query); } sqlFreeResult(&sr); hFreeConn(&conn); } void repeatMethods(struct track *tg) /* Make track for repeats. */ { tg->loadItems = repeatLoad; tg->freeItems = repeatFree; tg->drawItems = repeatDraw; tg->colorShades = shadesOfGray; tg->itemName = repeatName; tg->mapItemName = repeatName; tg->totalHeight = tgFixedTotalHeightNoOverflow; tg->itemHeight = tgFixedItemHeight; tg->itemStart = tgItemNoStart; tg->itemEnd = tgItemNoEnd; tg->mapsSelf = TRUE; }