Merge pull request #120 from smithlabcode/msite-input-validation

Msite input validation

Author: masarunakajima

src/analysis/hmr-rep.cpp

@@ -415,6 +415,10 @@ main_hmr_rep(int argc, const char **argv) {
     const vector<string> cpgs_files(leftover_args);
     /****************** END COMMAND LINE OPTIONS *****************/
 
+    for (auto &filename : cpgs_files)
+      if (!is_msite_file(filename))
+        throw runtime_error("malformed counts file: " + filename);
+
     vector<string> params_in_file;
     if (!params_in_files.empty()) {
       params_in_file = split_comma(params_in_files);

src/analysis/hmr.cpp

@@ -442,6 +442,9 @@ main_hmr(int argc, const char **argv) {
     const string cpgs_file = leftover_args.front();
     /****************** END COMMAND LINE OPTIONS *****************/
 
+    if (!is_msite_file(cpgs_file))
+      throw runtime_error("malformed counts file: " + cpgs_file);
+
     // separate the regions by chrom and by desert
     vector<MSite> cpgs;
     vector<pair<double, double> > meth;

src/analysis/hypermr.cpp

@@ -278,6 +278,9 @@ main_hypermr(int argc, const char **argv) {
     const string cpgs_file = leftover_args.front();
     /****************** END COMMAND LINE OPTIONS *****************/
 
+    if (!is_msite_file(cpgs_file))
+      throw runtime_error("malformed counts file: " + cpgs_file);
+
     if (VERBOSE) cerr << "[loading_data]" << endl;
     vector<MSite> cpgs;
     vector<pair<double, double>> meth;

src/analysis/levels.cpp

@@ -120,6 +120,9 @@ main_levels(int argc, const char **argv) {
     const string meth_file = leftover_args.front();
     /****************** END COMMAND LINE OPTIONS *****************/
 
+    if (!is_msite_file(meth_file))
+      throw runtime_error("malformed counts file: " + meth_file);
+
     const counts_file_format guessed_format =
       guess_counts_file_format(meth_file);
     if (guessed_format != counts_file_format::ordinary) {

src/analysis/pmd.cpp

@@ -1173,7 +1173,10 @@ main_pmd(int argc, const char **argv) {
       assert(cpgs_file.size() == params_in_file.size());
     }
 
-    size_t n_replicates = cpgs_file.size();
+    const size_t n_replicates = cpgs_file.size();
+    for (auto &filename : cpgs_file)
+      if (!is_msite_file(filename))
+        throw runtime_error("malformed counts file: " + filename);
 
     bool insufficient_data = false; // ADS: this is used now to detect
                                     // when the counts files have

src/analysis/roimethstat.cpp

@@ -470,6 +470,9 @@ Columns (beyond the first 6) in the BED format output:
     const string cpgs_file = leftover_args.back();
     /****************** END COMMAND LINE OPTIONS *****************/
 
+    if (!is_msite_file(cpgs_file))
+      throw runtime_error("malformed counts file: " + cpgs_file);
+
     if (VERBOSE)
       cerr << "loading regions" << endl;
 

src/common/MSite.cpp

@@ -156,43 +156,47 @@ find_offset_for_msite(const std::string &chr,
 }
 
 
-
 bool
-is_msite_file(const string &file) {
-  ifstream in(file);
-  if (!in)
-    throw runtime_error("cannot open file: " + file);
-
-  string line;
-  if(!getline(in, line)) return false;
+is_msite_line(const string &line) {
 
   std::istringstream iss(line);
 
   string chrom;
   if (!(iss >> chrom)) return false;
-  
+
   long int pos = 0;
   if (!(iss >> pos)) return false;
 
   string strand;
-  if (!(iss >> strand) || 
-      (strand.size() != 1) || 
-      ((strand != "+") && (strand != "-")) ) 
+  if (!(iss >> strand) ||
+      (strand.size() != 1) ||
+      ((strand != "+") && (strand != "-")) )
     return false;
 
   string context;
   std::regex pattern("^C[pHWX][GH]$");
   if (!(iss >> context) || !regex_match(context, pattern)) return false;
 
   double level = 0.0;
-  if (!(iss >> level) || level < 0 || level > 1) return false;
+  if (!(iss >> level) || level < 0.0 || level > 1.0) return false;
 
   long int n_reads = 0;
-  if (!(iss >> n_reads)) return false;
+  if (!(iss >> n_reads) || n_reads < 0) return false;
 
   string temp;
   if (iss >> temp) return false;
   else return true;
 
 }
 
+bool
+is_msite_file(const string &file) {
+  bamxx::bgzf_file in(file, "r");
+  if (!in)
+    throw runtime_error("cannot open file: " + file);
+
+  string line;
+  if (!getline(in, line)) return false;
+
+  return is_msite_line(line);
+}