Add proximity parameter and fix a bug in pairaln

Author: martin-steinegger

src/commons/Parameters.cpp

@@ -287,7 +287,9 @@ Parameters::Parameters():
         // pairaln
         PARAM_PAIRING_DUMMY_MODE(PARAM_PAIRING_DUMMY_MODE_ID, "--pairing-dummy-mode", "Include dummy pairing", "0: dont include, 1: include - an entry that will cause result2msa to write a gap only line", typeid(int), (void *) &pairdummymode, "^[0-1]{1}$", MMseqsParameter::COMMAND_EXPERT),
         PARAM_PAIRING_MODE(PARAM_PAIRING_MODE_ID, "--pairing-mode", "Pairing mode", "0: pair maximal per species, 1: pair only if all chains are covered per species", typeid(int), (void *) &pairmode, "^[0-1]{1}$", MMseqsParameter::COMMAND_EXPERT),
-        PARAM_PAIRING_FILTER(PARAM_PAIRING_FILTER_ID, "--pairing-filter", "Pairing filter", "Filter hits by 0: top hit, 1: pair by proximity of IDs", typeid(int), (void *) &pairfilter, "^[0-1]{1}$", MMseqsParameter::COMMAND_EXPERT),
+        PARAM_PAIRING_FILTER(PARAM_PAIRING_FILTER_ID, "--pairing-filter", "Pairing filter", "filter hits by 0: top hit, 1: pair by proximity of IDs", typeid(int), (void *) &pairfilter, "^[0-1]{1}$", MMseqsParameter::COMMAND_EXPERT),
+        PARAM_PAIRING_PROX_DISTANCE(PARAM_PAIRING_PROX_DISTANCE_ID, "--pairing-prox-distance", "Proximity distance threshold", "adjust the distance threshold for pairing (--pairing-filter 1)", typeid(int), (void *) &pairProximityDistance, "^[0-9]{1}[0-9]*$", MMseqsParameter::COMMAND_EXPERT),
+
         // taxonomyreport
         PARAM_REPORT_MODE(PARAM_REPORT_MODE_ID, "--report-mode", "Report mode", "Taxonomy report mode\n0: Kraken\n1: Krona\n2: do not create a report (for workflows only)\n3: Kraken per query database", typeid(int), (void *) &reportMode, "^[0-3]{1}$"),
         // createtaxdb
@@ -2681,6 +2683,7 @@ void Parameters::setDefaults() {
     pairdummymode = PAIRALN_DUMMY_MODE_OFF;
     pairmode = PAIRALN_MODE_ALL_PER_SPECIES;
     pairfilter = PAIRALN_FILTER_TOP_HIT;
+    pairProximityDistance = 20;
     // taxonomyreport
     reportMode = 0;
 

src/commons/Parameters.h

@@ -700,6 +700,7 @@ class Parameters {
     int pairdummymode;
     int pairmode;
     int pairfilter;
+    int pairProximityDistance;
 
     // taxonomyreport
     int reportMode;
@@ -1065,6 +1066,7 @@ class Parameters {
     PARAMETER(PARAM_PAIRING_DUMMY_MODE)
     PARAMETER(PARAM_PAIRING_MODE)
     PARAMETER(PARAM_PAIRING_FILTER)
+    PARAMETER(PARAM_PAIRING_PROX_DISTANCE)
 
     // taxonomyreport
     PARAMETER(PARAM_REPORT_MODE)

src/util/pairaln.cpp

@@ -357,8 +357,7 @@ int pairaln(int argc, const char **argv, const Command& command) {
                                 if (compatible[j].dbKey == UINT_MAX) continue;   // not set yet
                                 size_t prevNum = CompareUniProt::getUniProtNumber(compatible[j]);
                                 size_t diff = ABS_DIFF(currNum, prevNum);
-                                //TODO par.minPairDistance
-                                if (diff <= 20) {          // #4  adjust comparison as needed
+                                if (static_cast<int>(diff) <= par.pairProximityDistance) {
                                     isCompatible = true;
                                     break;
                                 }
@@ -377,6 +376,10 @@ int pairaln(int argc, const char **argv, const Command& command) {
                             bestCompatibleSize = compatibleSize;
                         }
                     }
+                    if(par.pairmode == Parameters::PAIRALN_MODE_COVER_ALL_CHAINS &&
+                            bestCompatibleSize != static_cast<int>(resultPerId.size())) {
+                        continue;
+                    }
                     for (size_t i = 0; i < bestCompatible.size(); i++) {
                         if (bestCompatible[i].dbKey == UINT_MAX &&
                             par.pairdummymode != Parameters::PAIRALN_DUMMY_MODE_ON) {