[FEATURE] add possibility to truncate subject ids on index creation

Author: h-2

src/lambda_indexer.hpp

@@ -53,13 +53,22 @@ inline int
 loadSubjSeqsAndIds(TCDStringSet<String<TOrigAlph>> & originalSeqs,
                    LambdaIndexerOptions const & options)
 {
-    StringSet<CharString, Owner<ConcatDirect<>>> ids;
+    typedef TCDStringSet<String<char, Alloc<>>>             TIDs;
+    typedef TCDStringSet<String<char, Alloc<Truncate_>>>    TIDsTruncated;
+    TIDs ids;
+    // difference only in name, same layout in mem, so we can hack the type:
+    TIDsTruncated* tIds = static_cast<TIDsTruncated*>((void*)&ids);
 
     double start = sysTime();
     myPrint(options, 1, "Loading Subject Sequences and Ids...");
 
     SeqFileIn infile(toCString(options.dbFile));
-    int ret = myReadRecords(ids, originalSeqs, infile);
+    int ret;
+    if (options.truncateIDs)
+        ret = myReadRecords(*tIds, originalSeqs, infile);
+    else
+        ret = myReadRecords(ids, originalSeqs, infile);
+
     if (ret)
         return ret;
 

src/misc.hpp

@@ -180,6 +180,54 @@ localAlignment2(Gaps<TSource0, TGapsSpec0> & row0,
     return score;
 }
 
+// ----------------------------------------------------------------------------
+// Tag Truncate_ (private tag to signify truncating of IDs while reading)
+// ----------------------------------------------------------------------------
+
+struct Truncate_;
+
+// ----------------------------------------------------------------------------
+// Function readRecord(Fasta); an overload that truncates Ids at first Whitespace
+// ----------------------------------------------------------------------------
+
+template <typename TSeqStringSet, typename TSpec, typename TSize>
+inline void
+readRecords(TCDStringSet<String<char, Alloc<Truncate_>>> & meta,
+            TSeqStringSet & seq,
+            FormattedFile<Fastq, Input, TSpec> & file,
+            TSize maxRecords)
+{
+    typedef typename SeqFileBuffer_<TSeqStringSet, TSpec>::Type TSeqBuffer;
+
+    TSeqBuffer seqBuffer;
+    IsWhitespace func;
+
+    // reuse the memory of context(file).buffer for seqBuffer (which has a different type but same sizeof(Alphabet))
+    swapPtr(seqBuffer.data_begin, context(file).buffer[1].data_begin);
+    swapPtr(seqBuffer.data_end, context(file).buffer[1].data_end);
+    seqBuffer.data_capacity = context(file).buffer[1].data_capacity;
+
+    for (; !atEnd(file) && maxRecords > 0; --maxRecords)
+    {
+        readRecord(context(file).buffer[0], seqBuffer, file);
+        for (size_t i = 0; i < length(context(file).buffer[0]); ++i)
+        {
+            if (func(context(file).buffer[0][i]))
+            {
+                resize(context(file).buffer[0], i);
+                break;
+            }
+        }
+        appendValue(meta, context(file).buffer[0]);
+        appendValue(seq, seqBuffer);
+    }
+
+    swapPtr(seqBuffer.data_begin, context(file).buffer[1].data_begin);
+    swapPtr(seqBuffer.data_end, context(file).buffer[1].data_end);
+    context(file).buffer[1].data_capacity = seqBuffer.data_capacity;
+    seqBuffer.data_capacity = 0;
+}
+
 // ----------------------------------------------------------------------------
 // Generic Sequence loading
 // ----------------------------------------------------------------------------

src/options.hpp

@@ -277,6 +277,8 @@ struct LambdaIndexerOptions : public SharedOptions
     std::string     segFile = "";
     std::string     algo = "";
 
+    bool            truncateIDs;
+
     LambdaIndexerOptions()
         : SharedOptions()
     {}
@@ -986,13 +988,21 @@ parseCommandLine(LambdaIndexerOptions & options, int argc, char const ** argv)
 //     setValidValues(parser, "output", "sa fm");
 
     addOption(parser, ArgParseOption("di", "db-index-type",
-        "suffix array or full-text minute space.",
+        "Suffix array or full-text minute space.",
         ArgParseArgument::STRING,
         "type"));
     setValidValues(parser, "db-index-type", "sa fm");
     setDefaultValue(parser, "db-index-type", "fm");
     setAdvanced(parser, "db-index-type");
 
+    addOption(parser, ArgParseOption("", "truncate-ids",
+        "Truncate IDs at first whitespace. This saves a lot of space and is irrelevant for all LAMBDA output formats "
+        "other than BLAST Pairwise (.m0).",
+        ArgParseArgument::STRING,
+        "STR"));
+    setValidValues(parser, "truncate-ids", "on off");
+    setDefaultValue(parser, "truncate-ids", "off");
+
     addSection(parser, "Alphabets and Translation");
     addOption(parser, ArgParseOption("p", "program",
         "Blast Operation Mode.",
@@ -1111,6 +1121,10 @@ parseCommandLine(LambdaIndexerOptions & options, int argc, char const ** argv)
     getOptionValue(tmpdir, parser, "tmp-dir");
     setEnv("TMPDIR", tmpdir);
 
+    std::string buffer;
+    getOptionValue(buffer, parser, "truncate-ids");
+    options.truncateIDs = (buffer == "on");
+
     return ArgumentParser::PARSE_OK;
 }