[FEATURE] expose macro to use 32bit as position type for proteins

seqan · Nov 23, 2016 · dd800f9 · dd800f9
1 parent 2d508a1
commit dd800f9
Show file tree

Hide file tree

Showing 5 changed files with 58 additions and 16 deletions.
diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt
@@ -83,11 +83,12 @@ message ("\n${ColourBold}Build configuration${ColourReset}")
 
 message (STATUS "LAMBDA version is: ${SEQAN_APP_VERSION}")
 
-option (LAMBDA_FASTBUILD    "Build only blastp and blastx modes (speeds up build)."                             OFF)
-option (LAMBDA_NATIVE_BUILD "Architecture-specific optimizations, i.e. g++ -march=native."                      ON)
-option (LAMBDA_STATIC_BUILD "Include all libraries in the binaries."                                            OFF)
-option (LAMBDA_MMAPPED_DB   "Use mmapped access to the database."                                               OFF)
-option (LAMBDA_LINGAPS_OPT  "Add optimized codepaths for linear gap costs (inc. bin size and compile time)."    OFF)
+option (LAMBDA_FASTBUILD                "Build only blastp and blastx modes (speeds up build)."                                     OFF)
+option (LAMBDA_LINGAPS_OPT              "Add optimized codepaths for linear gap costs (increases bin size and compile time)."       OFF)
+option (LAMBDA_LONG_PROTEIN_SUBJ_SEQS   "Make max protein sequence length == 4.3billion instead of 65,535. INVALIDATES INDEXS!"     OFF)
+option (LAMBDA_MMAPPED_DB               "Use mmapped access to the database."                                                       OFF)
+option (LAMBDA_NATIVE_BUILD             "Architecture-specific optimizations, i.e. g++ -march=native."                              ON)
+option (LAMBDA_STATIC_BUILD             "Include all libraries in the binaries."                                                    OFF)
 
 if (LAMBDA_FASTBUILD)
     add_definitions (-DFASTBUILD=1)
@@ -125,12 +126,17 @@ if (LAMBDA_LINGAPS_OPT)
     add_definitions (-DLAMBDA_LINGAPS_OPT=1)
 endif ()
 
+if (LAMBDA_LONG_PROTEIN_SUBJ_SEQS)
+    add_definitions (-DLAMBDA_LONG_PROTEIN_SUBJ_SEQS=1)
+endif ()
+
 message(STATUS "The following options are selected for the build:")
-message(   "     LAMBDA_FASTBUILD         ${LAMBDA_FASTBUILD}")
-message(   "     LAMBDA_LINGAPS_OPT       ${LAMBDA_LINGAPS_OPT}")
-message(   "     LAMBDA_MMAPPED_DB        ${LAMBDA_MMAPPED_DB}")
-message(   "     LAMBDA_NATIVE_BUILD      ${LAMBDA_NATIVE_BUILD}")
-message(   "     LAMBDA_STATIC_BUILD      ${LAMBDA_STATIC_BUILD}")
+message(   "     LAMBDA_FASTBUILD                ${LAMBDA_FASTBUILD}")
+message(   "     LAMBDA_LINGAPS_OPT              ${LAMBDA_LINGAPS_OPT}")
+message(   "     LAMBDA_LONG_PROTEIN_SUBJ_SEQS   ${LAMBDA_LONG_PROTEIN_SUBJ_SEQS}")
+message(   "     LAMBDA_MMAPPED_DB               ${LAMBDA_MMAPPED_DB}")
+message(   "     LAMBDA_NATIVE_BUILD             ${LAMBDA_NATIVE_BUILD}")
+message(   "     LAMBDA_STATIC_BUILD             ${LAMBDA_STATIC_BUILD}")
 message(STATUS "Run 'cmake -LH' to get a comment on each option.")
 message(STATUS "Remove CMakeCache.txt and re-run cmake with -DOPTIONNAME=ON|OFF to change an option.")
 

diff --git a/src/lambda.hpp b/src/lambda.hpp
@@ -128,7 +128,15 @@ template <typename TRedAlph,
 inline int
 validateIndexOptions(LambdaOptions const & options)
 {
-    //TODO verify that index dir exists
+    // Check that directory exists and is readable
+    struct stat path_stat;
+    stat(toCString(options.indexDir), &path_stat);
+    if (stat(toCString(options.indexDir), &path_stat) || !S_ISDIR(path_stat.st_mode))
+    {
+        std::cerr << "ERROR: Index directory does not exist or is not readable.\n";
+        return -1;
+    }
+
     std::string buffer;
     readIndexOption(buffer, "alph_translated", options);
     if (buffer != _alphName(TransAlph<p>()))
@@ -160,13 +168,30 @@ validateIndexOptions(LambdaOptions const & options)
     {
         buffer.clear();
         readIndexOption(buffer, "genetic_code", options);
-        unsigned long b = 0;
+        b = 0;
         if ((!lexicalCast(b, buffer)) || (b != static_cast<unsigned long>(options.geneticCode)))
         {
             std::cerr << "WARNING: The codon translation table used during indexing and during search are different. "
                          "This is not a problem per se, but is likely not what you want.\n\n";
         }
     }
+
+    buffer.clear();
+    readIndexOption(buffer, "subj_seq_len_bits", options);
+    b = 0;
+    if ((!lexicalCast(b, buffer)) || (b != static_cast<unsigned long>(sizeof(SizeTypePos_<TRedAlph>) * 8)))
+    {
+        #ifndef LAMBDA_LONG_PROTEIN_SUBJ_SEQS
+        std::cerr << "ERROR: Your lambda executable was built with LAMBDA_LONG_PROTEIN_SUBJ_SEQS,\n"
+                        "       but the index was created by an executable that was built without it.\n";
+        #else
+        std::cerr << "ERROR: Your lambda executable was built without LAMBDA_LONG_PROTEIN_SUBJ_SEQS,\n"
+                        "       but the index was created by an executable that was built with it.\n";
+        #endif
+        std::cerr << "       You need to recreate the index or rebuild Lambda.\n";
+        return -1;
+    }
+
     return 0;
 }
 

diff --git a/src/lambda_indexer.cpp b/src/lambda_indexer.cpp
@@ -197,7 +197,7 @@ realMain(LambdaIndexerOptions     const & options,
         dumpTranslatedSeqs(translatedSeqs, options);
 
     // see if final sequence set actually fits into index 
-    if (!checkIndexSize(translatedSeqs))
+    if (!checkIndexSize(translatedSeqs, BlastProgramSelector<p>()))
         return -1;
 
     if (options.dbIndexType == DbIndexType::FM_INDEX)
@@ -246,7 +246,8 @@ realMain(LambdaIndexerOptions     const & options,
              { options.indexDir + "/option:alph_original",   std::string(_alphName(OrigSubjAlph<p>())) },
              { options.indexDir + "/option:alph_translated", std::string(_alphName(TransAlph<p>())) },
              { options.indexDir + "/option:alph_reduced",    std::string(_alphName(TRedAlph())) },
-             { options.indexDir + "/option:genetic_code",    std::to_string(options.geneticCode) }
+             { options.indexDir + "/option:genetic_code",    std::to_string(options.geneticCode) },
+             { options.indexDir + "/option:subj_seq_len_bits", std::to_string(sizeof(SizeTypePos_<TRedAlph>) * 8)},
          })
     {
         std::ofstream f{std::get<0>(s).c_str(),  std::ios_base::out | std::ios_base::binary};

diff --git a/src/lambda_indexer.hpp b/src/lambda_indexer.hpp
@@ -276,9 +276,9 @@ dumpTranslatedSeqs(TCDStringSet<String<TTransAlph>> const & translatedSeqs,
 // Function loadSubj()
 // --------------------------------------------------------------------------
 
-template <typename TRedAlph>
+template <typename TRedAlph, BlastProgram p>
 inline bool
-checkIndexSize(TCDStringSet<String<TRedAlph>> const & seqs)
+checkIndexSize(TCDStringSet<String<TRedAlph>> const & seqs, BlastProgramSelector<p> const &)
 {
     using SAV = typename SAValue<TCDStringSet<String<TRedAlph>>>::Type;
     uint64_t curNumSeq = length(seqs);
@@ -303,6 +303,12 @@ checkIndexSize(TCDStringSet<String<TRedAlph>> const & seqs)
         std::cerr << "Too long sequences to be indexed:\n  "
                   << "length" << maxLen << " present in file, but only "
                   << maxLenSeq << " supported by index.\n";
+        #ifndef LAMBDA_LONG_PROTEIN_SUBJ_SEQS
+        if (p != BlastProgram::BLASTN)
+            std::cout << "You can recompile Lambda and add -DLAMBDA_LONG_PROTEIN_SUBJ_SEQS=1 to activate\n"
+                         "support for longer protein sequences.\n";
+        #endif
+
         return false;
     }
     return true;

diff --git a/src/options.hpp b/src/options.hpp
@@ -54,7 +54,11 @@ using SizeTypeNum_ = uint32_t;
 template <typename T>
 struct SizeTypePosMeta_
 {
+#ifdef LAMBDA_LONG_PROTEIN_SUBJ_SEQS
+    using Type = uint32_t;
+#else
     using Type = uint16_t;
+#endif
 };
 
 template <>