void checkLine(int lineNumber, char *line, char *regex, int childNum) {
ArrayList* result = matchingPositions(line, regex);
for (int i = 0; i < result->size; i++) {
int position = alGet(result, i);
printf("%d:%d\n", lineNumber, position);
// for (int k = 0; k < position; k++) printf(" ");
// printf("^\n");
}
alDelete(result);
usleep(100);
}
void mergeAll()
/* read through nameHash */
/* look up in 4 population hashes */
/* call logMissing to report on missing data */
/* check chrom, position, strand, observed */
{
char outputFileName[64];
FILE *outputFileHandle = NULL;
struct hashCookie cookie;
struct hashEl *nameHashElement = NULL;
struct hashEl *helCEU = NULL;
struct hashEl *helCHB = NULL;
struct hashEl *helJPT = NULL;
struct hashEl *helYRI = NULL;
boolean allMatch = TRUE;
struct hapmapSnpsCombined *hap = NULL;
safef(outputFileName, sizeof(outputFileName), "hapmapSnpsCombined.tab");
outputFileHandle = mustOpen(outputFileName, "w");
cookie = hashFirst(nameHash);
while ((nameHashElement = hashNext(&cookie)) != NULL)
{
helCEU = hashLookup(hashCEU, nameHashElement->name);
helCHB = hashLookup(hashCHB, nameHashElement->name);
helJPT = hashLookup(hashJPT, nameHashElement->name);
helYRI = hashLookup(hashYRI, nameHashElement->name);
/* should convert to instance of hapmapSnps here */
logMissing(nameHashElement->name, helCEU, helCHB, helJPT, helYRI);
allMatch = matchingChroms(helCEU, helCHB, helJPT, helYRI);
if (!allMatch)
{
fprintf(errorFileHandle, "different chroms for %s\n", nameHashElement->name);
continue;
}
allMatch = matchingPositions(helCEU, helCHB, helJPT, helYRI);
if (!allMatch)
{
fprintf(errorFileHandle, "different positions for %s\n", nameHashElement->name);
continue;
}
/* if strand is different, log and fix if possible */
allMatch = matchingStrand(helCEU, helCHB, helJPT, helYRI);
if (!allMatch)
{
fprintf(errorFileHandle, "different strands for %s\n", nameHashElement->name);
hap = fixStrandAndMerge(helCEU, helCHB, helJPT, helYRI);
if (hap)
writeOutput(hap, outputFileHandle);
continue;
}
/* different observed is not a fatal error? */
allMatch = matchingObserved(helCEU, helCHB, helJPT, helYRI);
if (!allMatch)
{
fprintf(errorFileHandle, "different observed for %s\n", nameHashElement->name);
continue;
}
/* should just have 2 alleles in all */
allMatch = matchingAlleles(helCEU, helCHB, helJPT, helYRI);
if (!allMatch)
{
fprintf(errorFileHandle, "different alleles for %s\n", nameHashElement->name);
continue;
}
hap = mergeOne(helCEU, helCHB, helJPT, helYRI);
if (hap)
writeOutput(hap, outputFileHandle);
}
carefulClose(&outputFileHandle);
}
