fasta::SequenceList fasta::remove_gaps(
const fasta::SequenceList& sequences) {
auto s = sequences;
for (auto& seq : s) {
seq.residues.clear();
}
for (size_t i = 0; i < sequences.size(); ++i) {
for (size_t j = 0; j < sequences[i].residues.size(); ++j) {
if (sequences[i].residues[j].codon[0] != '-') {
s[i].residues.push_back(sequences[i].residues[j]);
}
}
}
return s;
}
bool fasta::check_length(fasta::SequenceList const& sequences, int limit) {
if (limit == 0) {
limit = sequences.size();
}
auto result = true;
auto prev_length = sequences[0].residues.size();
auto i = 1;
while (result && i < limit) {
auto length = sequences[i].residues.size();
if (length != prev_length) {
result = false;
}
++i;
}
return result;
}
std::unordered_map<std::string, Occurences> FeatureScores::update_occurences(
const fasta::SequenceList& sequences,
const std::vector<double>& identities, const bool fade_out) {
std::unordered_map<std::string, Occurences> p;
for (auto& f: m_features) {
p[f] = std::vector<double>(sequences[0].residues.size(), 0);
}
for (size_t i = 0; i < sequences[0].residues.size(); ++i) {
for (size_t j = 0; j < sequences.size(); ++j) {
for (auto& f : sequences[j].residues[i].features) {
assert(p.find(f) != p.end());
if (fade_out) {
p[f][i] += 0.5 * (1 + identities[j]);
} else {
p[f][i] += 1.0;
}
}
}
}
return p;
}
void FeatureScores::update_scores(const fasta::SequenceList& sequences,
const f_config::FeatureSettingsMap& f_set,
const std::vector<double>& identities,
const bool fade_out) {
m_occurences = update_occurences(sequences, identities, fade_out);
// convert occurences to probabilities
for (auto& occ: m_occurences) {
size_t i = 0;
for (auto& v: occ.second) {
occ.second[i] = v / sequences.size();
++i;
}
}
// create an empty score map
std::unordered_map<std::string, Scores> scores;
for (auto& f: m_features) {
scores[f] = std::vector<double>(sequences[0].residues.size(), 0);
}
for (size_t i = 0; i < sequences[0].residues.size(); ++i) {
for (auto &feat : m_features){
if (feat.substr(0, 2) == "p_") {
scores[feat][i] = score_ptm(i, feat);
} else if (feat.substr(0, 2) == "d_") {
scores[feat][i] = score_domain(i, feat);
} else if (feat.substr(0, 2) == "m_") {
scores[feat][i] = score_motif(i, feat);
} else if (feat.substr(0, 2) == "s_") {
scores[feat][i] = score_strct(i, feat);
}
else if (feat.substr(0, 3) == "USR") {
scores[feat][i] = score_usr_feature(i, feat, f_set.at(feat));
}
}
}
m_scores = scores;
}