void addCloneInfo(char *glFileName, struct hash *cloneHash, struct clonePos **pCloneList)
/* Add in clone info from one .gl file. */
{
char dir[256], chrom[128], ext[64];
struct gl gl;
struct lineFile *lf = lineFileOpen(glFileName, TRUE);
struct clonePos *clone;
char *line, *words[8];
int lineSize, wordCount;
char cloneVerName[128];
char cloneName[128];
printf("Processing %s\n", glFileName);
splitPath(glFileName, dir, chrom, ext);
while (lineFileNext(lf, &line, &lineSize))
{
wordCount = chopLine(line, words);
if (wordCount == 0)
continue;
if (wordCount != 4)
errAbort("Expecting %d words line %d of %s", 4, lf->lineIx, lf->fileName);
glStaticLoad(words, &gl);
fragToCloneVerName(gl.frag, cloneVerName);
fragToCloneName(gl.frag, cloneName);
if ((clone = hashFindVal(cloneHash, cloneName)) == NULL)
{
AllocVar(clone);
clone->name = cloneString(cloneVerName);
clone->chrom = cloneString(chrom);
clone->chromStart = gl.start;
clone->chromEnd = gl.end;
clone->faFile = cloneString("");
slAddHead(pCloneList, clone);
hashAdd(cloneHash, cloneName, clone);
}
else
{
if (!sameString(clone->chrom, chrom))
{
warn("Clone %s is on chromosomes %s and %s. Ignoring %s",
cloneName, clone->chrom, chrom, chrom);
continue;
}
if (clone->chromStart > gl.start)
clone->chromStart = gl.start;
if (clone->chromEnd < gl.end)
clone->chromEnd = gl.end;
}
}
lineFileClose(&lf);
}
void addStretchInfo(char *fileName, struct chromInfo *ctg)
/* Add info about how much clones are stretched from gl file. */
{
struct hash *cloneHash = newHash(12);
struct lineFile *lf = lineFileOpen(fileName, TRUE);
char *row[4];
struct gl gl;
struct clone *cloneList = NULL, *clone;
while (lineFileRow(lf, row))
{
glStaticLoad(row, &gl);
chopSuffix(gl.frag);
if ((clone = hashFindVal(cloneHash, gl.frag)) == NULL)
{
AllocVar(clone);
slAddHead(&cloneList, clone);
hashAddSaveName(cloneHash, gl.frag, clone, &clone->name);
clone->start = gl.start;
clone->end = gl.end;
}
else
{
if (gl.start < clone->start) clone->start = gl.start;
if (gl.end > clone->end) clone->end = gl.end;
}
clone->totalSize += gl.end - gl.start;
}
for (clone = cloneList; clone != NULL; clone = clone->next)
{
int stretchSize = clone->end - clone->start;
double stretchRatio;
ctg->cloneCount += 1;
ctg->totalStretch += stretchSize - clone->totalSize;
stretchRatio = stretchSize / clone->totalSize;
if (stretchRatio > 1.3)
ctg->stretchedClones += 1;
if (stretchRatio > 2.0)
ctg->wayStretchedClones += 1;
}
lineFileClose(&lf);
hashFree(&cloneHash);
slFreeList(&cloneList);
}
void clonePosTab(char *fileName, struct hash *cloneHash)
/* Write out clonePos.tab. */
{
char query[256];
struct sqlResult *sr;
char **row;
struct clonePos *posList = NULL, *pos;
struct gl gl;
struct cloneInfo *info;
struct sqlConnection *conn = hAllocConn();
sprintf(query, "select * from chr18_gl");
sr = sqlGetResult(conn, query);
while ((row = sqlNextRow(sr)) != NULL)
{
glStaticLoad(row, &gl);
fragNameToCloneName(gl.frag);
info = findClone(cloneHash, gl.frag);
if ((pos = info->pos) == NULL)
{
AllocVar(pos);
pos->info = info;
info->pos = pos;
pos->start = gl.start;
pos->end = gl.end;
slAddHead(&posList, pos);
}
else
{
if (pos->start > gl.start)
pos->start = gl.start;
if (pos->end < gl.end)
pos->end = gl.end;
}
}
sqlFreeResult(&sr);
hFreeConn(&conn);
slSort(&posList, cmpClonePos);
writePosList(fileName, posList, "chr18");
}
