stop sorting the input mapping file

waveygang · Jun 1, 2024 · c738c1d · c738c1d
1 parent 70e896a
commit c738c1d
Showing 2 changed files with 4 additions and 71 deletions.
diff --git a/src/interface/main.cpp b/src/interface/main.cpp
@@ -91,76 +91,7 @@ int main(int argc, char** argv) {
         if (yeet_parameters.approx_mapping) {
             return 0;
         }
-     } else {
-        robin_hood::unordered_flat_map< std::string, std::pair<skch::seqno_t, uint64_t> > seqName_to_seqCounterAndLen;
-        skch::seqno_t seqCounter = 0;
-        for(const auto &fileName : map_parameters.querySequences) {
-            // check if there is a .fai
-            std::string fai_name = fileName + ".fai";
-            if (fs::exists(fai_name)) {
-                // if so, process the .fai to determine our sequence length
-                std::string line;
-                std::ifstream in(fai_name.c_str());
-                while (std::getline(in, line)) {
-                    auto line_split = skch::CommonFunc::split(line, '\t');
-                    const std::string seq_name = line_split[0];
-                    const uint64_t seq_len = std::stoull(line_split[1]);
-                    seqName_to_seqCounterAndLen[seq_name] = std::make_pair(seqCounter++,  seq_len);
-                }
-            } else {
-                // if not, warn that this is expensive
-                std::cerr << "[wfmash::align] WARNING, no .fai index found for " << fileName << ", reading the file to sort the mappings (slow)" << std::endl;
-                for(const auto &fileName : map_parameters.querySequences)
-                {
-                    seqiter::for_each_seq_in_file(
-						    fileName, {}, "", 
-                            [&](const std::string& seq_name,
-                                    const std::string& seq) {
-                                seqName_to_seqCounterAndLen[seq_name] = std::make_pair(seqCounter++,  seq.length());
-                            });
-                }
-            }
-        }
-
-
-        igzstream mappingListStream(map_parameters.outFileName.c_str());
-        std::string mappingRecordLine;
-        align::MappingBoundaryRow currentRecord;
-        std::vector<align::MappingBoundaryRow> allReadMappings;
-
-        while (!mappingListStream.eof()){
-            std::getline(mappingListStream, mappingRecordLine);
-            if( !mappingRecordLine.empty() ) {
-                align::Aligner::parseMashmapRow(mappingRecordLine, currentRecord);
-
-                allReadMappings.push_back(currentRecord);
-            }
-        }
-
-        std::sort(allReadMappings.begin(), allReadMappings.end(), [&seqName_to_seqCounterAndLen](const align::MappingBoundaryRow &a, const align::MappingBoundaryRow &b)
-        {
-            return (seqName_to_seqCounterAndLen[a.qId].first < seqName_to_seqCounterAndLen[b.qId].first);
-        });
-
-        std::ofstream outstrm(align_parameters.mashmapPafFile);
-        for(auto &e : allReadMappings)
-        {
-            outstrm << e.qId
-            << "\t" << seqName_to_seqCounterAndLen[e.qId].second
-            << "\t" << e.qStartPos
-            << "\t" << e.qEndPos
-            << "\t" << (e.strand == skch::strnd::FWD ? "+" : "-")
-            << "\t" << e.refId
-            << "\t" << seqName_to_seqCounterAndLen[e.refId].second
-            << "\t" << e.rStartPos
-            << "\t" << e.rEndPos
-            << "\t" << 0
-            << "\t" << std::max(e.rEndPos - e.rStartPos, e.qEndPos - e.qStartPos)
-            << "\t" << 255
-            << "\t" << "id:f:" << e.mashmap_estimated_identity
-            << "\n";
-        }
-    }
+     }
 
     if (align_parameters.sam_format) {
         // Prepare SAM header

diff --git a/src/interface/parse_args.hpp b/src/interface/parse_args.hpp
@@ -638,10 +638,12 @@ void parse_args(int argc,
         }
 
         if (align_input_paf) {
+            // directly use the input mapping file
             yeet_parameters.remapping = true;
             map_parameters.outFileName = args::get(align_input_paf);
-            align_parameters.mashmapPafFile = temp_file::create();
+            align_parameters.mashmapPafFile = args::get(align_input_paf);
         } else {
+            // make a temporary mapping file
             map_parameters.outFileName = temp_file::create();
             align_parameters.mashmapPafFile = map_parameters.outFileName;
         }