static
void handle_counts ( SOURCE* s,
CacheProfFile* cpf,
char* fi, char* fn, char* newCountsStr )
{
WordFM* countsMap;
Bool freeNewCounts;
UWord lnno;
Counts* newCounts;
FileFn* topKey;
if (0) printf("%s %s %s\n", fi, fn, newCountsStr );
// parse the numbers
newCounts = splitUpCountsLine( s, &lnno, newCountsStr );
// Did we get the right number?
if (newCounts->n_counts != cpf->n_events)
goto oom;
// allocate the key
topKey = malloc(sizeof(FileFn));
if (topKey) {
topKey->fi_name = strdup(fi);
topKey->fn_name = strdup(fn);
}
if (! (topKey && topKey->fi_name && topKey->fn_name))
mallocFail(s, "handle_counts:");
// search for it
if (lookupFM( cpf->outerMap, (Word*)(&countsMap), (Word)topKey )) {
// found it. Merge in new counts
freeNewCounts = addCountsToMap( s, countsMap, lnno, newCounts );
ddel_FileFn(topKey);
} else {
// not found in the top map. Create new entry
countsMap = newFM( malloc, free, cmp_unboxed_UWord );
if (!countsMap)
goto oom;
addToFM( cpf->outerMap, (Word)topKey, (Word)countsMap );
freeNewCounts = addCountsToMap( s, countsMap, lnno, newCounts );
}
// also add to running summary total
addCounts( s, cpf->summary, newCounts );
// if safe to do so, free up the count vector
if (freeNewCounts)
ddel_Counts(newCounts);
return;
oom:
parseError(s, "# counts doesn't match # events");
}
void mafStats(char *twoBitFile, char *mafDir, char *outFile)
/* mafStats - Calculate basic stats on maf file including species-by-species
* coverage and percent ID. */
{
struct twoBitFile *tbf = twoBitOpen(twoBitFile);
FILE *f = mustOpen(outFile, "w");
struct twoBitIndex *ix;
long genomeSize = 0;
struct hash *speciesHash = hashNew(0);
struct speciesAcc *speciesList = NULL, *species;
for (ix = tbf->indexList; ix != NULL; ix = ix->next)
{
unsigned chromSize = twoBitSeqSizeNoNs(tbf, ix->name);
genomeSize += chromSize;
char mafFileName[PATH_LEN];
safef(mafFileName, sizeof(mafFileName), "%s/%s.maf", mafDir, ix->name);
struct mafFile *mf = mafMayOpen(mafFileName);
verbose(1, "processing %s\n", ix->name);
if (mf == NULL)
{
warn("%s doesn't exist", mafFileName);
continue;
}
struct mafAli *maf;
while ((maf = mafNext(mf)) != NULL)
{
struct mafComp *mc;
for (mc = maf->components; mc != NULL; mc = mc->next)
{
if (mc->text != NULL)
toUpperN(mc->text, maf->textSize);
}
addCounts(maf, speciesHash, &speciesList);
mafAliFree(&maf);
}
mafFileFree(&mf);
}
slReverse(&speciesList);
for (species = speciesList; species != NULL; species = species->next)
{
fprintf(f, "counts: %s\t%ld\t%ld\t%ld\n", species->name, species->covCount, species->aliCount, species->idCount);
fprintf(f, "precents: %s\t%4.2f%%\t%4.2f%%\t%4.2f%%\n",
species->name, 100.0 * species->covCount/genomeSize,
100.0 * species->aliCount/genomeSize,
100.0 * species->idCount/species->aliCount);
}
carefulClose(&f);
}
static Bool addCountsToMap ( SOURCE* s,
WordFM* counts_map,
UWord lnno, Counts* newCounts )
{
Counts* oldCounts;
// look up lnno in the map. If none present, add a binding
// lnno->counts. If present, add counts to the existing entry.
if (lookupFM( counts_map, (Word*)(&oldCounts), (Word)lnno )) {
// merge with existing binding
addCounts( s, oldCounts, newCounts );
return True;
} else {
// create new binding
addToFM( counts_map, (Word)lnno, (Word)newCounts );
return False;
}
}
static void merge_CacheProfInfo ( SOURCE* s,
/*MOD*/CacheProfFile* dst,
CacheProfFile* src )
{
/* For each (filefn, innerMap) in src
if filefn not in dst
add binding dopy(filefn)->dopy(innerMap) in src
else
// merge src->innerMap with dst->innerMap
for each (lineno, counts) in src->innerMap
if lineno not in dst->innerMap
add binding lineno->dopy(counts) to dst->innerMap
else
add counts into dst->innerMap[lineno]
*/
/* Outer iterator: FileFn* -> WordFM* (inner iterator)
Inner iterator: UWord -> Counts*
*/
FileFn* soKey;
WordFM* soVal;
WordFM* doVal;
UWord siKey;
Counts* siVal;
Counts* diVal;
/* First check mundane things: that the events: lines are
identical. */
if (!streq( dst->events_line, src->events_line ))
barf(s, "\"events:\" line of most recent file does "
"not match those previously processed");
initIterFM( src->outerMap );
// for (filefn, innerMap) in src
while (nextIterFM( src->outerMap, (Word*)&soKey, (Word*)&soVal )) {
// is filefn in dst?
if (! lookupFM( dst->outerMap, (Word*)&doVal, (Word)soKey )) {
// no .. add dopy(filefn) -> dopy(innerMap) to src
FileFn* c_soKey = dopy_FileFn(soKey);
WordFM* c_soVal = dopy_InnerMap(soVal);
if ((!c_soKey) || (!c_soVal)) goto oom;
addToFM( dst->outerMap, (Word)c_soKey, (Word)c_soVal );
} else {
// yes .. merge the two innermaps
initIterFM( soVal );
// for (lno, counts) in soVal (source inner map)
while (nextIterFM( soVal, (Word*)&siKey, (Word*)&siVal )) {
// is lno in the corresponding dst inner map?
if (! lookupFM( doVal, (Word*)&diVal, siKey )) {
// no .. add lineno->dopy(counts) to dst inner map
Counts* c_siVal = dopy_Counts( siVal );
if (!c_siVal) goto oom;
addToFM( doVal, siKey, (Word)c_siVal );
} else {
// yes .. merge counts into dst inner map val
addCounts( s, diVal, siVal );
}
}
}
}
// add the summaries too
addCounts(s, dst->summary, src->summary );
return;
oom:
mallocFail(s, "merge_CacheProfInfo");
}
