static stList *readMatching(FILE *fileHandle, stList *originalEdges) {
/*
* Reads the matching created by Blossum.
*/
stHash *originalEdgesHash = putEdgesInHash(originalEdges);
char *line = stFile_getLineFromFile(fileHandle);
assert(line != NULL);
int64_t nodeNumber, edgeNumber;
int64_t i = sscanf(line, "%" PRIi64 " %" PRIi64 "\n", &nodeNumber, &edgeNumber);
assert(i == 2);
free(line);
stList *chosenEdges = stList_construct();
for(int64_t j=0; j<edgeNumber; j++) {
line = stFile_getLineFromFile(fileHandle);
int64_t node1, node2;
i = sscanf(line, "%" PRIi64 " %" PRIi64 "", &node1, &node2);
assert(i == 2);
free(line);
assert(node1 >= 0);
assert(node1 < nodeNumber);
assert(node2 >= 0);
assert(node2 < nodeNumber);
stIntTuple *edge = constructEdge(node1, node2);
stIntTuple *originalEdge = stHash_search(originalEdgesHash, edge);
if(originalEdge != NULL) {
stList_append(chosenEdges, originalEdge);
}
stIntTuple_destruct(edge);
}
stHash_destruct(originalEdgesHash);
return chosenEdges;
}
Hmm *hmm_loadFromFile(const char *fileName) {
FILE *fH = fopen(fileName, "r");
char *string = stFile_getLineFromFile(fH);
stList *tokens = stString_split(string);
if (stList_length(tokens) < 2) {
st_errAbort("Got an empty line in the input state machine file %s\n", fileName);
}
int type;
int64_t j = sscanf(stList_get(tokens, 0), "%i", &type);
if (j != 1) {
st_errAbort("Failed to parse state number (int) from string: %s\n", string);
}
Hmm *hmm = hmm_constructEmpty(0.0, type);
if (stList_length(tokens) != hmm->stateNumber * hmm->stateNumber + 2) {
st_errAbort(
"Got the wrong number of transitions in the input state machine file %s, got %" PRIi64 " instead of %" PRIi64 "\n",
fileName, stList_length(tokens), hmm->stateNumber * hmm->stateNumber + 2);
}
for (int64_t i = 0; i < hmm->stateNumber * hmm->stateNumber; i++) {
j = sscanf(stList_get(tokens, i + 1), "%lf", &(hmm->transitions[i]));
if (j != 1) {
st_errAbort("Failed to parse transition prob (float) from string: %s\n", string);
}
}
j = sscanf(stList_get(tokens, stList_length(tokens) - 1), "%lf", &(hmm->likelihood));
if (j != 1) {
st_errAbort("Failed to parse likelihood (float) from string: %s\n", string);
}
//Cleanup transitions line
free(string);
stList_destruct(tokens);
//Now parse the emissions line
string = stFile_getLineFromFile(fH);
tokens = stString_split(string);
if (stList_length(tokens) != hmm->stateNumber * SYMBOL_NUMBER_NO_N * SYMBOL_NUMBER_NO_N) {
st_errAbort(
"Got the wrong number of emissions in the input state machine file %s, got %" PRIi64 " instead of %" PRIi64 "\n",
fileName, stList_length(tokens), hmm->stateNumber * SYMBOL_NUMBER_NO_N * SYMBOL_NUMBER_NO_N);
}
for (int64_t i = 0; i < hmm->stateNumber * SYMBOL_NUMBER_NO_N * SYMBOL_NUMBER_NO_N; i++) {
j = sscanf(stList_get(tokens, i), "%lf", &(hmm->emissions[i]));
if (j != 1) {
st_errAbort("Failed to parse emission prob (float) from string: %s\n", string);
}
}
//Final cleanup
free(string);
stList_destruct(tokens);
fclose(fH);
return hmm;
}
stList *lineTokensFromFile(const char *filePath, int64_t getLine) {
FILE *fH = fopen(filePath, "r");
int64_t lineCount = 1;
stList *tokens;
char *string = stFile_getLineFromFile(fH);
while (string != NULL) {
string = stFile_getLineFromFile(fH);
if (lineCount == getLine) {
tokens = stString_split(string);
return tokens;
} else {
lineCount++;
free(string);
}
}
fclose(fH);
return stList_construct3(0, &free);
}
stSortedSet *loadEndAlignmentFromDisk(Flower *flower, FILE *fileHandle, End **end) {
stSortedSet *endAlignment =
stSortedSet_construct3((int (*)(const void *, const void *))alignedPair_cmpFn,
(void (*)(void *))alignedPair_destruct);
char *line = stFile_getLineFromFile(fileHandle);
if(line == NULL) {
*end = NULL;
return NULL;
}
Name flowerName;
int64_t lineNumber;
int64_t i = sscanf(line, "%" PRIi64 " %" PRIi64 "", &flowerName, &lineNumber);
if(i != 2 || lineNumber < 0) {
st_errAbort("We encountered a mis-specified name in loading the first line of an end alignment from the disk: '%s'\n", line);
}
free(line);
*end = flower_getEnd(flower, flowerName);
if(*end == NULL) {
st_errAbort("We encountered an end name that is not in the database: '%s'\n", line);
}
for(int64_t i=0; i<lineNumber; i++) {
line = stFile_getLineFromFile(fileHandle);
if(line == NULL) {
st_errAbort("Got a null line when parsing an end alignment\n");
}
int64_t sI1, sI2;
int64_t p1, st1, p2, st2, score1, score2;
int64_t i = sscanf(line, "%" PRIi64 " %" PRIi64 " %" PRIi64 " %" PRIi64 " %" PRIi64 " %" PRIi64 " %" PRIi64 " %" PRIi64 "", &sI1, &p1, &st1, &score1, &sI2, &p2, &st2, &score2);
(void)i;
if(i != 8) {
st_errAbort("We encountered a mis-specified name in loading an end alignment from the disk: '%s'\n", line);
}
stSortedSet_insert(endAlignment, alignedPair_construct(sI1, p1, st1, sI2, p2, st2, score1, score2));
free(line);
}
return endAlignment;
}