void
verify_node_orig(kg_node_t * node, unsigned kmer_length) {
assert( false && "TODO FIX! REVERSED KMER ENDIANNESS" );
int double_kmer_length = kmer_length << 1;
#ifdef LARGE_KMERS
Kmer mask;
mask.createMask(double_kmer_length);
#else
Kmer mask = (Kmer(1) << double_kmer_length) - 1;
#endif
Kmer kmer = node->kmer;
Kmer rc_kmer = reverseComplement(kmer, kmer_length);
char leftmost_base = (kmer >> (double_kmer_length - 2)) & 0x3;
char rightmost_base = kmer & 0x3;
for (int i = 0 ; i < 4 ; ++ i) {
// check on the left side
kg_node_t * node2 = node->left[i];
int count = node->left_count[i];
if (node2) {
assert (count != 0);
if (count > 0) {
Kmer kmer2 = KMER_PREPEND(kmer, i, double_kmer_length, mask);
assert(kmer2 == node2->kmer);
assert(node2->right[(int)rightmost_base] == node);
assert(node2->right_count[(int)rightmost_base] == count);
} else {
Kmer kmer2 = KMER_APPEND(rc_kmer, i ^ 0x3, double_kmer_length, mask);
assert(kmer2 == node2->kmer);
assert(node2->left[rightmost_base ^ 0x3] == node);
assert(node2->left_count[rightmost_base ^ 0x3] == count);
}
} else {
assert (count == 0);
}
// check on the right side
node2 = node->right[i];
count = node->right_count[i];
if (node2) {
assert (count != 0);
if (count > 0) {
Kmer kmer2 = KMER_APPEND(kmer, i, double_kmer_length, mask);
assert(kmer2 == node2->kmer);
assert(node2->left[(int)leftmost_base] == node);
assert(node2->left_count[(int)leftmost_base] == count);
} else {
Kmer kmer2 = KMER_PREPEND(rc_kmer, i ^ 0x3, double_kmer_length, mask);
assert(kmer2 == node2->kmer);
assert(node2->right[leftmost_base ^ 0x3] == node);
assert(node2->right_count[leftmost_base ^ 0x3] == count);
}
} else {
assert (count == 0);
}
}
}
void printTranscripts(const KmerIndex& index, std::stringstream& o, const std::string s,
const std::vector<std::pair<KmerEntry,int>>& v, const std::vector<int> u) {
Kmer km;
KmerEntry val;
int p;
// find first mapping k-mer
if (!v.empty()) {
p = findFirstMappingKmer(v,val);
km = Kmer((s.c_str()+p));
}
for (int i = 0; i < u.size(); i++) {
int tr = u[i];
if (i > 0) {
o << ";";
}
std::pair<int, bool> xp = index.findPosition(tr, km, val, p);
o << "(" << index.target_names_[tr] << "," << xp.first << ",";
if (xp.second) {
o << "FW)";
} else {
o << "RC)";
}
}
}
void DeleteKMers(const Sequence &nucls, EdgeId e) {
VERIFY(nucls.size() >= index_.k());
KeyWithHash kwh = index_.ConstructKWH(Kmer(index_.k(), nucls));
DeleteIfEqual(kwh, e);
for (size_t i = index_.k(), n = nucls.size(); i < n; ++i) {
kwh <<= nucls[i];
DeleteIfEqual(kwh, e);
}
}
void verify_node(KmerNode * node, KmerGraph *hashtable, unsigned kmer_length)
{
int double_kmer_length = kmer_length << 1;
#ifdef LARGE_KMERS
Kmer mask;
mask.createMask(double_kmer_length);
#else
Kmer mask = (Kmer(1) << double_kmer_length) - 1;
#endif
Kmer kmer = node->kmer;
Kmer rc_kmer = reverseComplement(kmer, kmer_length);
char rightmost_base = KMER_GET_TAIL_BASE(kmer, kmer_length);
char leftmost_base = KMER_GET_HEAD_BASE(kmer, kmer_length);
KmerNode *node2;
for (int i = 0 ; i < 4 ; ++ i) {
// check on the left side
int count = node->left_count[i];
int color = node->left_color[i];
assert( color == 0 || count != 0 ); // count must be non-zero if color is non-zero
if (color == 0) {
if (count > 0) {
Kmer kmer2 = KMER_PREPEND(kmer, i, double_kmer_length, mask);
node2 = hashtable->findNode(canonicalKmer(kmer2, kmer_length));
assert( node2 != NULL );
assert(cnorm(node2->right_count[static_cast<int>(rightmost_base)]) == cnorm(count));
assert(node2->right_color[static_cast<int>(rightmost_base)] == 0);
} else if (count < 0) {
Kmer kmer2 = KMER_APPEND(rc_kmer, i ^ 0x3, double_kmer_length);
node2 = hashtable->findNode(canonicalKmer(kmer2, kmer_length));
assert( node2 != NULL );
assert(cnorm(node2->left_count[static_cast<int>( COMPLEMENT(rightmost_base) )]) == cnorm(count));
assert(node2->left_color[static_cast<int>( COMPLEMENT(rightmost_base) )] == 0);
}
}
// check on the right side
count = node->right_count[i];
color = node->right_color[i];
assert( color == 0 || count != 0 ); // count must be non-zero if color is non-zero
if (color == 0) {
if (count > 0) {
Kmer kmer2 = KMER_APPEND(kmer, i, double_kmer_length);
node2 = hashtable->findNode(canonicalKmer(kmer2, kmer_length));
assert( node2 != NULL );
assert(cnorm(node2->left_count[static_cast<int>(leftmost_base)]) == cnorm(count));
assert(node2->left_color[static_cast<int>(leftmost_base)] == 0);
} else if (count < 0) {
Kmer kmer2 = KMER_PREPEND(rc_kmer, i ^ 0x3, double_kmer_length, mask);
node2 = hashtable->findNode(canonicalKmer(kmer2, kmer_length));
assert( node2 != NULL );
assert(cnorm(node2->right_count[static_cast<int>( COMPLEMENT(leftmost_base) )]) == cnorm(count));
assert(node2->right_color[static_cast<int>( COMPLEMENT(leftmost_base) )] == 0);
}
}
}
}
void testSimpleInsertACAutomaton() {
PointerACAutomaton<Kmer> aho;
Kmer count_acag = Kmer("ACAG");
count_acag.count = 5;
Kmer count_caga = Kmer("CAGA");
count_caga.count = 2;
Kmer count_caca = Kmer("CACA");
count_caca.count = 3;
Kmer count_gca = Kmer("GCA");
count_gca.count = 1;
aho.insert("ACAG", count_acag);
aho.insert("CAGA", count_caga);
aho.insert("CACA", count_caca);
aho.insert("GCA", count_gca);
aho.build_failure_functions();
pointer_state<Kmer> *state_ac = aho.goto_state("ac");
pointer_state<Kmer> *state_aca = aho.goto_state("aca");
pointer_state<Kmer> *state_c = aho.goto_state("c");
pointer_state<Kmer> *state_g = aho.goto_state("g");
pointer_state<Kmer> *state_cag = aho.goto_state("cag");
TAP_TEST(aho.getInitialState()->transitions[T] == aho.getInitialState()
&& aho.getInitialState()->transitions[N] == aho.getInitialState(),
TEST_AC_TRANSITIONS, "");
TAP_TEST(state_ac->transitions[A] == state_aca, TEST_AC_TRANSITIONS, "");
TAP_TEST(state_ac->transitions[C] == state_c, TEST_AC_TRANSITIONS, "");
TAP_TEST(state_aca->transitions[A] == state_g->transitions[A], TEST_AC_TRANSITIONS, "");
TAP_TEST(state_c->transitions[C] == state_c, TEST_AC_TRANSITIONS, "");
TAP_TEST(state_g->transitions[G] == state_g, TEST_AC_TRANSITIONS, "");
TAP_TEST(state_g->transitions[C]->transitions[C] == state_c, TEST_AC_TRANSITIONS, "");
TAP_TEST(state_cag->transitions[G] == state_g, TEST_AC_TRANSITIONS, "");
TAP_TEST(state_cag->transitions[A]->is_final, TEST_AC_FINAL, "");
string caga = "caga";
string caca = "caca";
string acag = "acag";
TAP_TEST(aho.get(caga).count == 2, TEST_AC_GET, "");
TAP_TEST(aho.get(caca).count == 3, TEST_AC_GET, "");
TAP_TEST(aho.get(acag).count == 5, TEST_AC_GET, "");
}
void dump_hash(const kmers_map_type& kmers_map, const char* prefix, const char* suffix) {
FILE* fout;
std::ostringstream fname;
fname << prefix << suffix << ".hash";
fout = fopen(fname.str().c_str(), "wb");
FILE* findexout;
std::ostringstream findexname;
findexname << prefix << suffix << ".index";
findexout = fopen(findexname.str().c_str(), "wb");
// compute index
std::vector<std::tr1::tuple<Kmer, const occ_list_type* > > index;
index.push_back(std::tr1::make_tuple(Kmer("Z"), (occ_list_type*) 0));
for (kmers_map_type::const_iterator it = kmers_map.begin(); it != kmers_map.end(); it++) {
index.push_back(std::tr1::make_tuple(it->first, &(it->second)));
}
// Remove the (Kmer("Z"),0) tuple.
index.erase(index.begin());
std::sort(index.begin(), index.end(), index_less());
// Write out index.
// Format: kmer string, number of reads with kmer
// Writes out:
// num kmers in block, (kmer, num reads with kmer)+
const unsigned int nKmer = index.size();
fwrite(&nKmer, sizeof(unsigned int), 1, findexout); // number of kmers
for (unsigned int kid = 0; kid < nKmer; kid++) { // index of kmer
const char* kmer = std::tr1::get<0>(index[kid]).getKmer().c_str();
unsigned int nread = std::tr1::get<1>(index[kid])->size();
fwrite(kmer, sizeof(char), strlen(kmer) + 1, findexout);
fwrite(&nread, sizeof(unsigned int), 1, findexout);
}
// Write out reads.
// Format: read ID, hit position
// Writes out:
// read id, pos within read
for (unsigned int kid = 0; kid < nKmer; kid++) {
const Kmer& kmer = std::tr1::get<0>(index[kid]);
const occ_list_type& occ_list = *std::tr1::get<1>(index[kid]);
for (occ_list_type::const_iterator it = occ_list.begin(); it != occ_list.end(); it++) {
const unsigned int& readid = it->readID;
const unsigned int& pos = it->pos;
fwrite(&readid, sizeof(unsigned int), 1, fout);
fwrite(&pos, sizeof(unsigned int), 1, fout);
}
}
fclose(fout);
fclose(findexout);
}
static CountDB fromFile( const std::string& fname ) {
std::ifstream in(fname, std::ios::in | std::ios::binary );
uint32_t merSize;
in.read( reinterpret_cast<char*>(&merSize), sizeof(merSize) );
size_t numCounts;
in.read( reinterpret_cast<char*>(&numCounts), sizeof(size_t) );
auto kmers = std::make_shared<std::vector<Kmer>>(numCounts, Kmer(0));
std::vector<AtomicCount> counts(numCounts);
in.read( reinterpret_cast<char*>(&((*kmers)[0])), sizeof(Kmer) * numCounts );
in.read( reinterpret_cast<char*>(&counts[0]), sizeof(AtomicCount) * numCounts );
in.close();
CountDB index(kmers, merSize);
index._counts = std::move(counts);
return index;
}
bool checkMapability(const KmerIndex& index, const std::string &s, const std::vector<std::pair<KmerEntry,int>>& v, std::vector<int> &u) {
const int maxMismatch = 2;
const int maxSoftclip = 5;
Kmer km;
KmerEntry val;
int p;
if (!v.empty()) {
p = findFirstMappingKmer(v,val);
km = Kmer(s.c_str()+p);
} else {
return false;
}
std::vector<int> vtmp; vtmp.reserve(u.size());
for (auto tr : u) {
auto trpos = index.findPosition(tr, km, val, p);
int tpos = (int)trpos.first;
int sz = (int)s.size();
bool add = true;
if (trpos.second) {
if (tpos < 1 || tpos + sz - 1 > index.target_seqs_[tr].size()) {
add = false;
} else {
//std::cout << index.target_seqs_[tr].substr(tpos,sz) << std::endl;
//std::cout << s << std::endl;
int mis = 0;
for (int i = 0; i < sz - maxSoftclip; i++) {
if (index.target_seqs_[tr][tpos-1 + i] != s[i]) {
++mis;
if (mis > maxMismatch) {
break;
}
}
}
add = (mis <= maxMismatch);
}
} else {
if (tpos > index.target_seqs_[tr].size() || tpos - sz < 1) {
add = false;
} else {
std::string rs = revcomp(s);
//std::cout << index.target_seqs_[tr].substr(tpos - sz, sz) << std::endl;
//std::cout << rs << std::endl;
int mis = 0;
for (int i = sz-1; i >= maxSoftclip; i--) {
if (index.target_seqs_[tr][tpos-sz+i] != rs[sz]) {
++mis;
if (mis > maxMismatch) {
break;
}
}
}
add = (mis <= maxMismatch);
}
}
if (add) {
vtmp.push_back(tr);
}
}
if (vtmp.empty()) {
return false;
}
if (vtmp.size() < u.size()) {
u = vtmp; // copy
}
return true;
}
