Load ncbi (6.1.20040616) into ncbi-tools6/branches/upstream/current.

9 files changed, 271 insertions, 569 deletions

diff --git a/demo/asn2fsa.c b/demo/asn2fsa.c
index a385ffb3..2cc68fdd 100644
--- a/demo/asn2fsa.c
+++ b/demo/asn2fsa.c
@@ -29,7 +29,7 @@
 *
 * Version Creation Date:   3/4/04
 *
-* $Revision: 1.13 $
+* $Revision: 1.16 $
 *
 * File Description:
 *
@@ -61,6 +61,10 @@
 #include <accpubseq.h>
 #endif
 
+#define ASN2FSA_APP_VER "1.0"
+
+CharPtr ASN2FSA_APPLICATION = ASN2FSA_APP_VER;
+
 static ValNodePtr  requested_uid_list = NULL;
 static TNlmMutex   requested_uid_mutex = NULL;
 
@@ -265,6 +269,7 @@ typedef struct fastaflags {
   FILE     *aa;
   FILE     *ql;
   FILE     *fr;
+  FILE     *logfp;
 } FastaFlagData, PNTR FastaFlagPtr;
 
 static VoidPtr DoAsyncLookup (
@@ -405,6 +410,9 @@ static ValNodePtr DoLockFarComponents (
 
   if (NlmThreadsAvailable () && ffp->useThreads) {
     rsult = AsyncLockFarComponents (sep, ffp);
+  } else if (ffp->useThreads) {
+    Message (MSG_POST, "Threads not available in this executable");
+    rsult = LockFarComponents (sep);
   } else {
     rsult = LockFarComponents (sep);
   }
@@ -673,13 +681,16 @@ static void ProcessMultipleRecord (
   AsnIoPtr       aip;
   AsnModulePtr   amp;
   AsnTypePtr     atp, atp_bss, atp_desc, atp_se;
+  BioseqPtr      bsp;
   ValNodePtr     bsplist;
-  Char           cmmd [256], file [FILENAME_MAX], path [PATH_MAX];
+  Char           buf [64], cmmd [256], file [FILENAME_MAX], path [PATH_MAX], longest [64];
   Char           path1 [PATH_MAX], path2 [PATH_MAX], path3 [PATH_MAX];
   StreamFlgType  flags = 0;
   FILE           *fp, *tfp;
+  Int4           numrecords = 0;
+  SeqEntryPtr    fsep, sep;
   ObjMgrPtr      omp;
-  SeqEntryPtr    sep;
+  time_t         starttime, stoptime, worsttime;
 #ifdef OS_UNIX
   CharPtr        gzcatprog;
   int            ret;
@@ -796,10 +807,28 @@ static void ProcessMultipleRecord (
     flags = STREAM_EXPAND_GAPS;
   }
 
+  longest [0] = '\0';
+  worsttime = 0;
+
   while ((atp = AsnReadId (aip, amp, atp)) != NULL) {
     if (atp == atp_se) {
       sep = SeqEntryAsnRead (aip, atp);
 
+      starttime = GetSecs ();
+      buf [0] = '\0';
+
+      if (ffp->logfp != NULL) {
+        fsep = FindNthBioseq (sep, 1);
+        if (fsep != NULL && fsep->choice == 1) {
+          bsp = (BioseqPtr) fsep->data.ptrvalue;
+          if (bsp != NULL) {
+            SeqIdWrite (bsp->id, buf, PRINTID_FASTA_LONG, sizeof (buf));
+            fprintf (ffp->logfp, "%s\n", buf);
+            fflush (ffp->logfp);
+          }
+        }
+      }
+
       bsplist = NULL;
       if (ffp->lock) {
         bsplist = DoLockFarComponents (sep, ffp);
@@ -820,6 +849,13 @@ static void ProcessMultipleRecord (
 
       bsplist = UnlockFarComponents (bsplist);
 
+      stoptime = GetSecs ();
+      if (stoptime - starttime > worsttime && StringDoesHaveText (buf)) {
+        worsttime = stoptime - starttime;
+        StringCpy (longest, buf);
+      }
+      numrecords++;
+
       SeqEntryFree (sep);
       omp = ObjMgrGet ();
       ObjMgrReapOne (omp);
@@ -841,6 +877,13 @@ static void ProcessMultipleRecord (
   FileClose (fp);
 #endif
 
+  if (ffp->logfp != NULL && (! StringHasNoText (longest))) {
+    fprintf (ffp->logfp, "Longest processing time %ld seconds on %s\n",
+             (long) worsttime, longest);
+    fprintf (ffp->logfp, "Total number of records %ld\n", (long) numrecords);
+    fflush (ffp->logfp);
+  }
+
   sprintf (cmmd, "rm %s; rm %s; rm %s", path1, path2, path3);
   system (cmmd);
 }
@@ -939,6 +982,7 @@ static void FileRecurse (
 #define h_argFarOutFile   20
 #define e_argLineLength   21
 #define T_argThreads      22
+#define L_argLogFile      23
 
 Args myargs [] = {
   {"Path to ASN.1 Files", NULL, NULL, NULL,
@@ -987,21 +1031,22 @@ Args myargs [] = {
     TRUE, 'e', ARG_INT, 0.0, 0, NULL},
   {"Use Threads", "F", NULL, NULL,
     TRUE, 'T', ARG_BOOLEAN, 0.0, 0, NULL},
+  {"Log File", NULL, NULL, NULL,
+    TRUE, 'L', ARG_FILE_OUT, 0.0, 0, NULL},
 };
 
 Int2 Main (void)
 
 {
+  Char           app [64], sfx [32];
   CharPtr        base, blastdb, directory, fastaidx, ntout,
-                 aaout, qlout, frout, ptr, str, suffix;
+                 aaout, qlout, frout, logfile, ptr, str, suffix;
   Boolean        batch, binary, blast, compressed, dorecurse,
                  expandgaps, fargenomicqual, fasta, local, lock,
                  masterstyle, qualgapzero, remote, usethreads;
   FastaFlagData  ffd;
-  FILE           *fp = NULL;
   Int2           linelen, type = 0;
   time_t         run_time, start_time, stop_time;
-  Char           sfx [32];
 
   /* standard setup */
 
@@ -1034,7 +1079,8 @@ Int2 Main (void)
 
   /* process command line arguments */
 
-  if (! GetArgs ("asn2fsa", sizeof (myargs) / sizeof (Args), myargs)) {
+  sprintf (app, "asn2fsa %s", ASN2FSA_APPLICATION);
+  if (! GetArgs (app, sizeof (myargs) / sizeof (Args), myargs)) {
     return 0;
   }
 
@@ -1054,7 +1100,7 @@ Int2 Main (void)
   local = (Boolean) myargs [k_argLocalFetch].intvalue;
   lock = (Boolean) myargs [l_argLockFar].intvalue;
   linelen = (Int2) myargs [e_argLineLength].intvalue;
-  usethreads = (Int2) myargs [T_argThreads].intvalue;
+  usethreads = (Boolean) myargs [T_argThreads].intvalue;
 
   expandgaps = (Boolean) myargs [g_argExpandGaps].intvalue;
   masterstyle = (Boolean) myargs [m_argMaster].intvalue;
@@ -1091,6 +1137,8 @@ Int2 Main (void)
   qlout = (CharPtr) myargs [q_argQlOutFile].strvalue;
   frout = (CharPtr) myargs [h_argFarOutFile].strvalue;
 
+  logfile = (CharPtr) myargs [L_argLogFile].strvalue;
+
   /* default to stdout for nucleotide output if nothing specified */
 
   if (StringHasNoText (ntout) &&
@@ -1118,6 +1166,7 @@ Int2 Main (void)
   ffd.aa = NULL;
   ffd.ql = NULL;
   ffd.fr = NULL;
+  ffd.logfp = NULL;
 
   if (! StringHasNoText (ntout)) {
     ffd.nt = FileOpen (ntout, "w");
@@ -1152,6 +1201,14 @@ Int2 Main (void)
     ffd.lock = TRUE;
   }
 
+  if (! StringHasNoText (logfile)) {
+    ffd.logfp = FileOpen (logfile, "w");
+    if (ffd.logfp == NULL) {
+      Message (MSG_FATAL, "Unable to open log file");
+      return 1;
+    }
+  }
+
   /* register fetch functions */
 
   if (remote) {
@@ -1161,6 +1218,7 @@ Int2 Main (void)
       return 1;
     }
     ffd.usePUBSEQ = TRUE;
+    ffd.useThreads = FALSE;
 #else
     PubSeqFetchEnable ();
 #endif
@@ -1227,6 +1285,12 @@ Int2 Main (void)
 
   stop_time = GetSecs ();
   run_time = stop_time - start_time;
+
+  if (ffd.logfp != NULL) {
+    fprintf (ffd.logfp, "Finished in %ld seconds\n", (long) run_time);
+    FileClose (ffd.logfp);
+  }
+
   Message (MSG_POST, "Ran in %ld seconds", (long) run_time);
 
   /* close fetch functions */
diff --git a/demo/asn2gb.c b/demo/asn2gb.c
index 0598a4bd..14411743 100644
--- a/demo/asn2gb.c
+++ b/demo/asn2gb.c
@@ -29,7 +29,7 @@
 *
 * Version Creation Date:   10/21/98
 *
-* $Revision: 6.61 $
+* $Revision: 6.64 $
 *
 * File Description:  New GenBank flatfile generator application
 *
@@ -50,6 +50,10 @@
 #include <explore.h>
 #include <asn2gnbp.h>
 
+#define ASN2GB_APP_VER "2.0"
+
+CharPtr ASN2GB_APPLICATION = ASN2GB_APP_VER;
+
 static void SaveSeqEntry (
   SeqEntryPtr sep,
   CharPtr filename
@@ -145,6 +149,7 @@ static Int2 HandleSingleRecord (
   CharPtr inputFile,
   CharPtr outputFile,
   FmtType format,
+  FmtType altformat,
   ModType mode,
   StlType style,
   FlgType flags,
@@ -323,6 +328,9 @@ static Int2 HandleSingleRecord (
         AsnIoFree (aip, FALSE);
       } else {
         SeqEntryToGnbk (sep, slp, format, mode, style, flags, locks, custom, extra, ofp);
+        if (altformat != 0) {
+          SeqEntryToGnbk (sep, slp, altformat, mode, style, flags, locks, custom, extra, ofp);
+        }
       }
       if (ofp != NULL) {
         FileClose (ofp);
@@ -469,6 +477,7 @@ static void CompareFlatFiles (
   SeqEntryPtr sep,
   FILE* fp,
   FmtType format,
+  FmtType altformat,
   ModType mode,
   StlType style,
   FlgType flags,
@@ -496,6 +505,9 @@ static void CompareFlatFiles (
   if (batch == 1) {
 
     SeqEntryToGnbk (sep, NULL, format, mode, style, flags, locks, custom, extra, fp);
+    if (altformat != 0) {
+      SeqEntryToGnbk (sep, NULL, altformat, mode, style, flags, locks, custom, extra, fp);
+    }
     return; /* just make report, nothing to diff */
 
   } else if (batch == 2) {
@@ -591,6 +603,9 @@ static void CompareFlatFiles (
 #else
 
   SeqEntryToGnbk (sep, NULL, format, mode, style, flags, locks, custom, extra, fp);
+  if (altformat != 0) {
+    SeqEntryToGnbk (sep, NULL, altformat, mode, style, flags, locks, custom, extra, fp);
+  }
 #endif
 }
 
@@ -648,6 +663,7 @@ static Int2 HandleMultipleRecords (
   CharPtr inputFile,
   CharPtr outputFile,
   FmtType format,
+  FmtType altformat,
   ModType mode,
   StlType style,
   FlgType flags,
@@ -883,8 +899,8 @@ static Int2 HandleMultipleRecords (
           starttime = GetSecs ();
           useGbdjoin = (Boolean) (format == GENBANK_FMT && (! hasRefSeq));
           CompareFlatFiles (path1, path2, path3, sep, ofp,
-                            format, mode, style, flags, locks, custom,
-                            extra, batch, gbdjoin, useGbdjoin);
+                            format, altformat, mode, style, flags, locks,
+                            custom, extra, batch, gbdjoin, useGbdjoin);
           stoptime = GetSecs ();
           if (stoptime - starttime > worsttime) {
             worsttime = stoptime - starttime;
@@ -968,7 +984,7 @@ Args myargs [] = {
     FALSE, 'i', ARG_FILE_IN, 0.0, 0, NULL},
   {"Output File Name", "stdout", NULL, NULL,
     FALSE, 'o', ARG_FILE_OUT, 0.0, 0, NULL},
-  {"Format (b GenBank, e EMBL, p GenPept, t Feature Table, x GBSet)", "b", NULL, NULL,
+  {"Format (b GenBank, e EMBL, p GenPept, t Feature Table, x INSDSet)", "b", NULL, NULL,
     FALSE, 'f', ARG_STRING, 0.0, 0, NULL},
   {"Mode (r Release, e Entrez, s Sequin, d Dump)", "s", NULL, NULL,
     FALSE, 'm', ARG_STRING, 0.0, 0, NULL},
@@ -1033,6 +1049,8 @@ Int2 Main (
 {
   CharPtr     accn = NULL;
   AsnIoPtr    aip = NULL;
+  FmtType     altformat = (FmtType) 0;
+  Char        app [64];
   AsnTypePtr  atp = NULL;
   Int2        batch = 0;
   Boolean     binary = FALSE;
@@ -1054,8 +1072,6 @@ Int2 Main (
   CharPtr     logfile = NULL;
   FILE        *logfp = NULL;
   ModType     mode = SEQUIN_MODE;
-  Char        path [PATH_MAX];
-  CharPtr     progname;
   Boolean     propOK = FALSE;
   Int2        rsult = 0;
   time_t      runtime, starttime, stoptime;
@@ -1096,17 +1112,10 @@ Int2 Main (
     return 1;
   }
 
-  ProgramPath (path, sizeof (path));
-  progname = StringRChr (path, DIRDELIMCHR);
-  if (progname != NULL) {
-    progname++;
-  } else {
-    progname = "asn2gb";
-  }
-
   /* process command line arguments */
 
-  if (! GetArgs (progname, sizeof (myargs) / sizeof (Args), myargs)) {
+  sprintf (app, "asn2gb %s", ASN2GB_APPLICATION);
+  if (! GetArgs (app, sizeof (myargs) / sizeof (Args), myargs)) {
     return 0;
   }
 
@@ -1131,7 +1140,11 @@ Int2 Main (
   }
 
   str = myargs [f_argFormat].strvalue;
-  if (StringICmp (str, "b") == 0) {
+  if (StringICmp (str, "bp") == 0 || StringICmp (str, "pb") == 0) {
+    format = GENBANK_FMT;
+    altformat = GENPEPT_FMT;
+
+  } else if (StringICmp (str, "b") == 0) {
     format = GENBANK_FMT;
   } else if (StringICmp (str, "e") == 0) {
     format = EMBL_FMT;
@@ -1304,13 +1317,13 @@ Int2 Main (
   if (batch != 0 || accn != NULL) {
     rsult = HandleMultipleRecords (myargs [i_argInputFile].strvalue,
                                    myargs [o_argOutputFile].strvalue,
-                                   format, mode, style, flags, locks,
+                                   format, altformat, mode, style, flags, locks,
                                    custom, extra, batch, binary, compressed,
                                    propOK, gbdjoin, accn, logfp);
   } else {
     rsult = HandleSingleRecord (myargs [i_argInputFile].strvalue,
                                 myargs [o_argOutputFile].strvalue,
-                                format, mode, style, flags, locks,
+                                format, altformat, mode, style, flags, locks,
                                 custom, extra, type, binary, compressed,
                                 from, to, strand, itemID, do_tiny_seq, do_fasta_stream);
   }
@@ -1322,12 +1335,11 @@ Int2 Main (
   }
 
   stoptime = GetSecs ();
+  runtime = stoptime - starttime;
   if (logfp != NULL) {
-    fprintf (logfp, "Finished in %ld seconds\n",
-             (long) (stoptime - starttime));
+    fprintf (logfp, "Finished in %ld seconds\n", (long) runtime);
     FileClose (logfp);
   }
-  runtime = stoptime - starttime;
   Message (MSG_POST, "Ran in %ld seconds", (long) runtime);
 
   if (myargs [r_argRemote].intvalue) {
diff --git a/demo/blast_driver.c b/demo/blast_driver.c
index f568f993..16a4bb01 100644
--- a/demo/blast_driver.c
+++ b/demo/blast_driver.c
@@ -1,4 +1,4 @@
-/* $Id: blast_driver.c,v 1.40 2004/05/05 15:30:33 dondosha Exp $
+/* $Id: blast_driver.c,v 1.46 2004/06/08 17:47:43 dondosha Exp $
 * ===========================================================================
 *
 *                            PUBLIC DOMAIN NOTICE
@@ -32,10 +32,10 @@ Author: Ilya Dondoshansky
 Contents: Main function for running BLAST
 
 ******************************************************************************
- * $Revision: 1.40 $
+ * $Revision: 1.46 $
  * */
 
-static char const rcsid[] = "$Id: blast_driver.c,v 1.40 2004/05/05 15:30:33 dondosha Exp $";
+static char const rcsid[] = "$Id: blast_driver.c,v 1.46 2004/06/08 17:47:43 dondosha Exp $";
 
 #include <ncbi.h>
 #include <sqnutils.h>
@@ -46,6 +46,8 @@ static char const rcsid[] = "$Id: blast_driver.c,v 1.40 2004/05/05 15:30:33 dond
 #include <algo/blast/core/blast_filter.h>
 #include <algo/blast/core/blast_util.h>
 #include <algo/blast/core/blast_engine.h>
+#include <algo/blast/core/hspstream_collector.h>
+#include <algo/blast/api/hspstream_queue.h>
 #include <algo/blast/api/blast_seq.h>
 #include <algo/blast/api/blast_input.h>
 #include <algo/blast/api/blast_format.h>
@@ -53,6 +55,7 @@ static char const rcsid[] = "$Id: blast_driver.c,v 1.40 2004/05/05 15:30:33 dond
 #include <algo/blast/api/blast_format.h>
 #include <algo/blast/api/seqsrc_readdb.h>
 #include <algo/blast/api/multiseq_src.h>
+#include <algo/blast/api/blast_tabular.h>
 
 #define NUMARG (sizeof(myargs)/sizeof(myargs[0]))
 
@@ -97,7 +100,8 @@ typedef enum {
    ARG_FORMAT,
    ARG_HTML,
    ARG_ASNOUT,
-   ARG_OIDRANGE
+   ARG_OIDRANGE,
+   ARG_TABULAR
 } BlastArguments;
 
 static Args myargs[] = {
@@ -199,7 +203,9 @@ static Args myargs[] = {
    { "Range of ordinal ids in the BLAST database to search.\n"
      "Format: \"oid1 oid2\"; ',', ':' or ';' can also be used as delimiters\n" 
      "Full database is searched if range not provided.", /* ARG_OIDRANGE */
-     NULL, NULL, NULL, TRUE, 'R', ARG_STRING, 0.0, 0, NULL}
+     NULL, NULL, NULL, TRUE, 'R', ARG_STRING, 0.0, 0, NULL},
+   { "Produce on-the-fly tabular output",
+     "0", NULL, NULL, FALSE, 'B', ARG_INT, 0.0, 0, NULL} /* ARG_TABULAR */
 };
 
 static Int2 BLAST_FillRPSInfo( RPSInfo **ppinfo, Nlm_MemMap **rps_mmap,
@@ -314,10 +320,9 @@ BLAST_FillOptions(LookupTableOptions* lookup_options,
 {
    char* blast_program;
    Boolean ag_blast = TRUE, variable_wordsize = FALSE, mb_lookup = FALSE;
-   Boolean greedy_extension = FALSE;
+   Int4 greedy_extension = 0;
+   Boolean greedy_with_ungapped = FALSE;
    Boolean is_gapped = FALSE;
-   Int8 totlen = 0;
-   Int4 numseqs = 0;
    Uint1 program_number;
    Int2 status;
    Boolean use_pssm = FALSE;
@@ -340,7 +345,8 @@ BLAST_FillOptions(LookupTableOptions* lookup_options,
          mb_lookup = TRUE;
          variable_wordsize = FALSE;
       }
-      greedy_extension = (Boolean) myargs[ARG_GREEDY].intvalue;
+      greedy_extension = MIN(myargs[ARG_GREEDY].intvalue, 2);
+      greedy_with_ungapped = (myargs[ARG_GREEDY].intvalue == 3);
    }
 
    BLAST_FillLookupTableOptions(lookup_options, program_number, mb_lookup,
@@ -372,42 +378,26 @@ BLAST_FillOptions(LookupTableOptions* lookup_options,
       query_setup_options->genetic_code = myargs[ARG_GENCODE].intvalue;
 
    BLAST_FillInitialWordOptions(word_options, program_number, 
-      greedy_extension, myargs[ARG_WINDOW].intvalue, variable_wordsize, 
-      ag_blast, mb_lookup, myargs[ARG_XDROP_UNGAPPED].intvalue);
+      (greedy_extension && !greedy_with_ungapped), 
+      myargs[ARG_WINDOW].intvalue, variable_wordsize, ag_blast, mb_lookup, 
+      myargs[ARG_XDROP_UNGAPPED].intvalue);
 
    BLAST_FillExtensionOptions(ext_options, program_number, greedy_extension, 
       myargs[ARG_XDROP].intvalue, myargs[ARG_XDROP_FINAL].intvalue);
 
-   if (greedy_extension) {
-       switch (myargs[ARG_GREEDY].intvalue) {
-       case 1:
-           ext_options->algorithm_type = EXTEND_GREEDY;
-           word_options->ungapped_extension = FALSE;
-           break;
-       case 2:
-           ext_options->algorithm_type = EXTEND_GREEDY_NO_TRACEBACK;
-           word_options->ungapped_extension = FALSE;
-           break;
-       case 3:
-           ext_options->algorithm_type = EXTEND_GREEDY_NO_TRACEBACK;
-           word_options->ungapped_extension = TRUE;
-           break;
-       default:
-           break;
-       }
-   }
-
    if (program_number == blast_type_rpsblast ||
-       program_number == blast_type_rpstblastn)
-      BLAST_FillScoringOptions(score_options, program_number, greedy_extension, 
+       program_number == blast_type_rpstblastn) {
+      BLAST_FillScoringOptions(score_options, program_number, FALSE,
                 myargs[ARG_MISMATCH].intvalue, myargs[ARG_MATCH].intvalue,
                 "BLOSUM62", rps_info->aux_info.gap_open_penalty,
                 rps_info->aux_info.gap_extend_penalty);
-   else
-      BLAST_FillScoringOptions(score_options, program_number, greedy_extension, 
+   } else {
+      BLAST_FillScoringOptions(score_options, program_number, 
+                (Boolean)greedy_extension, 
                 myargs[ARG_MISMATCH].intvalue, myargs[ARG_MATCH].intvalue,
                 myargs[ARG_MATRIX].strvalue, myargs[ARG_GAPOPEN].intvalue,
                 myargs[ARG_GAPEXT].intvalue);
+   }
 
    if (program_number != blast_type_tblastx)
       is_gapped = !myargs[ARG_UNGAPPED].intvalue;
@@ -426,8 +416,9 @@ BLAST_FillOptions(LookupTableOptions* lookup_options,
    hit_options->percent_identity = myargs[ARG_PERC_IDENT].floatvalue;
    hit_options->longest_intron = myargs[ARG_INTRON].intvalue;
 
-   BLAST_FillEffectiveLengthsOptions(eff_len_options, 
-      numseqs, totlen, (Int8) myargs[ARG_SEARCHSP].floatvalue);
+   if (myargs[ARG_SEARCHSP].floatvalue != 0) {
+      eff_len_options->searchsp_eff = (Int8) myargs[ARG_SEARCHSP].floatvalue; 
+   }
 
    if (db_options && (program_number == blast_type_tblastn ||
                       program_number == blast_type_rpstblastn ||
@@ -458,7 +449,7 @@ Int2 Nlm_Main(void)
    BlastHitSavingOptions* hit_options;
    char* dbname = NULL;
    LookupTableWrap* lookup_wrap;
-   Int2 status;
+   Int2 status = 0;
    QuerySetUpOptions* query_options=NULL;	
    BlastEffectiveLengthsOptions* eff_len_options=NULL;
    BlastMaskLoc* lcase_mask = NULL;
@@ -472,7 +463,7 @@ Int2 Nlm_Main(void)
    SeqAlign* seqalign;
    BlastFormattingOptions* format_options;
    Boolean done;
-   BlastReturnStat* return_stats;
+   BlastDiagnostics* diagnostics;
    Int4 ctr = 0;
    PSIBlastOptions* psi_options = NULL;
    BlastDatabaseOptions* db_options = NULL;
@@ -485,6 +476,11 @@ Int2 Nlm_Main(void)
    Nlm_MemMapPtr rps_mmap = NULL;
    Nlm_MemMapPtr rps_pssm_mmap = NULL;
    RPSInfo *rps_info = NULL;
+   double scale_factor;
+   BlastHSPStream* hsp_stream = NULL;
+   Boolean tabular_output;
+   TNlmThread format_thread;
+   BlastTabularFormatData* tf_data = NULL;
 
    if (! GetArgs (buf, NUMARG, myargs))
       return (1);
@@ -496,11 +492,13 @@ Int2 Nlm_Main(void)
    
    ErrSetMessageLevel(SEV_WARNING);
    
-   if ((outfp = fopen(myargs[ARG_OUT].strvalue, "w")) == NULL) {
+   if ((outfp = FileOpen(myargs[ARG_OUT].strvalue, "w")) == NULL) {
       ErrPostEx(SEV_FATAL, 1, 0, "blast: Unable to open output file %s\n", 
                 myargs[ARG_OUT].strvalue);
      return (1);
    }
+
+   tabular_output = (Boolean)myargs[ARG_TABULAR].intvalue;
    
    blast_program = strdup(myargs[ARG_PROGRAM].strvalue);
    BlastProgram2Number(myargs[ARG_PROGRAM].strvalue, &program_number);
@@ -525,7 +523,7 @@ Int2 Nlm_Main(void)
    if (!myargs[ARG_DB].strvalue) {
       FILE *infp2;
       char *subject_file = strdup(myargs[ARG_SUBJECT].strvalue);
-      if ((infp2 = fopen(subject_file, "r")) == NULL) {
+      if ((infp2 = FileOpen(subject_file, "r")) == NULL) {
          ErrPostEx(SEV_FATAL, 1, 0, 
                    "blast: Unable to open second input file %s\n", 
                    subject_file);
@@ -535,7 +533,7 @@ Int2 Nlm_Main(void)
 
       BLAST_GetQuerySeqLoc(infp2, db_is_na, 0, 0, 0, NULL, &subject_slp, 
                            0, NULL);
-      fclose(infp2);
+      FileClose(infp2);
       
       seq_src = MultiSeqSrcInit(subject_slp, program_number);
 
@@ -558,34 +556,39 @@ Int2 Nlm_Main(void)
       if (BLAST_FillRPSInfo(&rps_info, &rps_mmap, 
                             &rps_pssm_mmap, myargs[ARG_DB].strvalue) != 0)
          ErrPostEx(SEV_FATAL, 1, 0,  "RPS Blast setup failed");
+      scale_factor = rps_info->aux_info.scale_factor;
+   }
+   else {
+      scale_factor = 1.0;
    }
 
    BLAST_FillOptions(lookup_options, query_options, word_options, 
       ext_options, hit_options, score_options, eff_len_options, 
       psi_options, db_options, seq_src, rps_info);
+   if (!tabular_output) {
+      if ((status = BlastFormattingOptionsNew(program_number, 
+                       myargs[ARG_OUT].strvalue, 
+                       myargs[ARG_DESCRIPTIONS].intvalue, 
+                       myargs[ARG_ALIGNMENTS].intvalue, 
+                       myargs[ARG_FORMAT].intvalue, &format_options)) != 0)
+         return status;
+      format_options->html = (Boolean) myargs[ARG_HTML].intvalue;
 
-   if ((status = BlastFormattingOptionsNew(program_number, 
-                    myargs[ARG_OUT].strvalue, 
-                    myargs[ARG_DESCRIPTIONS].intvalue, 
-                    myargs[ARG_ALIGNMENTS].intvalue, 
-                    myargs[ARG_FORMAT].intvalue, &format_options)) != 0)
-      return status;
-   format_options->html = (Boolean) myargs[ARG_HTML].intvalue;
-
-   if (seq_src) {
-      dbname = BLASTSeqSrcGetName(seq_src);
+      if (seq_src) {
+         dbname = BLASTSeqSrcGetName(seq_src);
 
-      BLAST_PrintOutputHeader(format_options, 
-         myargs[ARG_GREEDY].intvalue, dbname, !db_is_na);
+         BLAST_PrintOutputHeader(format_options, 
+            myargs[ARG_GREEDY].intvalue, dbname, !db_is_na);
+      }
    }
 
-   if ((infp = fopen(myargs[ARG_QUERY].strvalue, "r")) == NULL) {
+   if ((infp = FileOpen(myargs[ARG_QUERY].strvalue, "r")) == NULL) {
       ErrPostEx(SEV_FATAL, 1, 0, "blast: Unable to open input file %s\n", 
                 myargs[ARG_QUERY].strvalue);
       return (1);
    }
    
-   return_stats = (BlastReturnStat*) calloc(1, sizeof(BlastReturnStat));
+   diagnostics = Blast_DiagnosticsInit();
 
    translated_query = (program_number == blast_type_blastx || 
                        program_number == blast_type_tblastx);
@@ -622,7 +625,7 @@ Int2 Nlm_Main(void)
 
       status = 
          BLAST_MainSetUp(program_number, query_options, score_options, 
-            hit_options, query, query_info, &lookup_segments, 
+            hit_options, query, query_info, scale_factor, &lookup_segments, 
             &filter_loc, &sbp, &blast_message);
 
       if (translated_query) {
@@ -638,42 +641,75 @@ Int2 Nlm_Main(void)
          return status;
       }
 
-      Blast_HSPResultsInit(query_info->num_queries, &results);
       LookupTableWrapInit(query, lookup_options, 
                           lookup_segments, sbp, &lookup_wrap, rps_info);
     
-      if (rps_blast)
+      if (!tabular_output) {
+         Int4 num_results = (rps_blast ? BLASTSeqSrcGetNumSeqs(seq_src) : 
+                             query_info->num_queries);
+         /* Results in the collector stream should be sorted only for a
+            database search. The latter is true if and only if the sequence
+            source has non-zero database length. */
+         Boolean sort_on_read = (BLASTSeqSrcGetTotLen(seq_src) != 0);
+         hsp_stream = 
+            Blast_HSPListCollectorInit(program_number, hit_options, 
+                                       num_results, sort_on_read);
+      } else {
+         hsp_stream = Blast_HSPListQueueInit();
+         tf_data = Blast_TabularFormatDataInit(program_number, hsp_stream, 
+                      seq_src, query, query_info, score_options, sbp, 
+                      eff_len_options, ext_options, hit_options, db_options, 
+                      query_slp, outfp);
+         /* Start the formatting thread */
+         if((format_thread = 
+             NlmThreadCreate(Blast_TabularFormatThread, (void*) tf_data))
+            == NULL_thread) {
+            fprintf(stderr, 
+                    "Cannot create thread for formatting tabular output\n");
+            return 1;
+         }
+      }
+
+      if (rps_blast) {
          BLAST_RPSSearchEngine(program_number, query, query_info, 
             seq_src, sbp, score_options, lookup_wrap, 
             word_options, ext_options, hit_options, eff_len_options, 
-            psi_options, db_options, results, return_stats);
-      else
+            psi_options, db_options, hsp_stream, diagnostics, 
+            (tabular_output ? NULL : &results));
+      } else {
          BLAST_SearchEngine(program_number, query, query_info, 
             seq_src, sbp, score_options, lookup_wrap, 
             word_options, ext_options, hit_options, eff_len_options, 
-            psi_options, db_options, results, return_stats);
+            psi_options, db_options, hsp_stream, diagnostics, 
+            (tabular_output ? NULL : &results));
+      }
 
+      if (tabular_output) {
+         void* join_status = NULL;
+         NlmThreadJoin(format_thread, &join_status);
+      }
+
+      hsp_stream = BlastHSPStreamFree(hsp_stream);
       lookup_wrap = LookupTableWrapFree(lookup_wrap);
 
       if (rps_blast) {
-	 Nlm_MemMapFini(rps_mmap);
-	 Nlm_MemMapFini(rps_pssm_mmap);
+         Nlm_MemMapFini(rps_mmap);
+         Nlm_MemMapFini(rps_pssm_mmap);
          sfree(rps_info->aux_info.karlin_k);
          sfree(rps_info->aux_info.orig_score_matrix);
-	 sfree(rps_info);
+         sfree(rps_info);
       }
 
       /* The following works because the ListNodes' data point to simple
          double-integer structures */
       lookup_segments = ListNodeFreeData(lookup_segments);
-
+      if (!tabular_output) {
       /* Convert results to the SeqAlign form */
       BLAST_ResultsToSeqAlign(program_number, results, query_slp, seq_src, 
-         subject_slp, score_options, sbp, score_options->gapped_calculation,
+         score_options->gapped_calculation, score_options->is_ooframe, 
          &seqalign);
 
       results = Blast_HSPResultsFree(results);
-      seq_src = BlastSeqSrcFree(seq_src);
       
       if (myargs[ARG_ASNOUT].strvalue) {
          AsnIoPtr asnout = AsnIoOpen(myargs[ARG_ASNOUT].strvalue, (char*)"w");
@@ -686,21 +722,21 @@ Int2 Nlm_Main(void)
       status = BLAST_FormatResults(seqalign, dbname, 
                   blast_program, query_info->num_queries, query_slp,
                   filter_loc, format_options, score_options->is_ooframe);
-
-      BlastMaskLocFree(filter_loc);
-
       PrintOutputFooter(program_number, format_options, score_options, sbp, 
-         lookup_options, word_options, ext_options, hit_options, query_info, 
-         dbname, return_stats, db_is_na);
-
+                        lookup_options, word_options, ext_options, 
+                        hit_options, eff_len_options, query_info, 
+                        seq_src, diagnostics);
+      } /* if not tabular output */
       query = BlastSequenceBlkFree(query);
+      BlastMaskLocFree(filter_loc);
       query_info = BlastQueryInfoFree(query_info);
       BlastScoreBlkFree(sbp);
       query_slp = SeqLocSetFree(query_slp);
    } /* End loop on sets of queries */
    
+   seq_src = BlastSeqSrcFree(seq_src);
    subject_slp = SeqLocSetFree(subject_slp);
-   sfree(return_stats);
+   Blast_DiagnosticsFree(diagnostics);
    LookupTableOptionsFree(lookup_options);
    BlastQuerySetUpOptionsFree(query_options);
    BlastExtensionOptionsFree(ext_options);
@@ -710,11 +746,14 @@ Int2 Nlm_Main(void)
    BlastEffectiveLengthsOptionsFree(eff_len_options);
    PSIBlastOptionsFree(psi_options);
    BlastDatabaseOptionsFree(db_options);
-   
-   BlastFormattingOptionsFree(format_options);
+   if (!tabular_output) { 
+      BlastFormattingOptionsFree(format_options);
+   } else {
+      FileClose(outfp);
+   }
 
    if (infp)
-      fclose(infp);
+      FileClose(infp);
    
    sfree(dbname);
    sfree(blast_program);
diff --git a/demo/blastall.c b/demo/blastall.c
index 0b712792..0ead6d9d 100644
--- a/demo/blastall.c
+++ b/demo/blastall.c
@@ -1,6 +1,6 @@
-static char const rcsid[] = "$Id: blastall.c,v 6.142 2004/04/29 19:56:00 dondosha Exp $";
+static char const rcsid[] = "$Id: blastall.c,v 6.143 2004/05/13 18:42:44 coulouri Exp $";
 
-/* $Id: blastall.c,v 6.142 2004/04/29 19:56:00 dondosha Exp $
+/* $Id: blastall.c,v 6.143 2004/05/13 18:42:44 coulouri Exp $
 **************************************************************************
 *                                                                         *
 *                             COPYRIGHT NOTICE                            *
@@ -28,6 +28,9 @@ static char const rcsid[] = "$Id: blastall.c,v 6.142 2004/04/29 19:56:00 dondosh
 ************************************************************************** 
  * 
  * $Log: blastall.c,v $
+ * Revision 6.143  2004/05/13 18:42:44  coulouri
+ * disable -B for blastcl3
+ *
  * Revision 6.142  2004/04/29 19:56:00  dondosha
  * Mask filtered locations in query sequence lines in XML output
  *
@@ -762,8 +765,10 @@ static Args myargs[] = {
     sizeof(myargs) itself
    made optional=TRUE but this may change?
 */
+#ifndef BLAST_CS_API
     { "Number of concatenated queries, for blastn and tblastn", /* 40 */
       "0", NULL, NULL, TRUE, 'B', ARG_INT, 0.0, 0, NULL}
+#endif
 };
 
 #ifdef BLAST_CS_API
diff --git a/demo/fastacmd.c b/demo/fastacmd.c
index 010bae9d..ff0521cc 100644
--- a/demo/fastacmd.c
+++ b/demo/fastacmd.c
@@ -1,6 +1,6 @@
-static char const rcsid[] = "$Id: fastacmd.c,v 6.29 2003/05/30 17:31:09 coulouri Exp $";
+static char const rcsid[] = "$Id: fastacmd.c,v 6.30 2004/05/13 20:54:45 coulouri Exp $";
 
-/* $Id: fastacmd.c,v 6.29 2003/05/30 17:31:09 coulouri Exp $
+/* $Id: fastacmd.c,v 6.30 2004/05/13 20:54:45 coulouri Exp $
 * ===========================================================================
 *
 *                            PUBLIC DOMAIN NOTICE
@@ -31,12 +31,15 @@ static char const rcsid[] = "$Id: fastacmd.c,v 6.29 2003/05/30 17:31:09 coulouri
 *
 * Initial Version Creation Date: 05/20/1997
 *
-* $Revision: 6.29 $
+* $Revision: 6.30 $
 *
 * File Description:
 *        FASTA retrievel system using ISAM indexes
 *
 * $Log: fastacmd.c,v $
+* Revision 6.30  2004/05/13 20:54:45  coulouri
+* spell 'loci' correctly
+*
 * Revision 6.29  2003/05/30 17:31:09  coulouri
 * add rcsid
 *
@@ -157,10 +160,10 @@ static Args myargs [] = {
       "         T - protein   \n"
       "         F - nucleotide", 
       "G", NULL,NULL,TRUE,'p',ARG_STRING,0.0,0,NULL},
-    { "Search string: GIs, accessions and locuses may be used delimited\n"
+    { "Search string: GIs, accessions and loci may be used delimited\n"
       "      by comma.",                                        /* 2 */
       NULL, NULL, NULL, TRUE, 's', ARG_STRING, 0.0, 0, NULL},
-    { "Input file wilth GIs/accessions/locuses for batch\n"
+    { "Input file wilth GIs/accessions/loci for batch\n"
       "      retrieval",/* 3 */
       NULL, NULL, NULL, TRUE, 'i', ARG_STRING, 0.0, 0, NULL},
     { "Retrieve duplicate accessions",                          /* 4 */
diff --git a/demo/fmerge.c b/demo/fmerge.c
deleted file mode 100644
index 1aaae442..00000000
--- a/demo/fmerge.c
+++ /dev/null
@@ -1,445 +0,0 @@
-/*****************************************************************************
-
-    Name: fmerge.c                               
-
-    Description: Program for merging FASTA databases.
-
-    Author: Sergei Shavirin
-
-   ***************************************************************************
-
-                          PUBLIC DOMAIN NOTICE
-              National Center for Biotechnology Information
-
-    This software/database is a "United States Government Work" under the
-    terms of the United States Copyright Act.  It was written as part of    
-    the author's official duties as a United States Government employee
-    and thus cannot be copyrighted.  This software/database is freely
-    available to the public for use. The National Library of Medicine and
-    the U.S. Government have not placed any restriction on its use or
-    reproduction.
-
-    Although all reasonable efforts have been taken to ensure the accuracy
-    and reliability of the software and data, the NLM and the U.S.
-    Government do not and cannot warrant the performance or results that
-    may be obtained by using this software or data. The NLM and the U.S.
-    Government disclaim all warranties, express or implied, including
-    warranties of performance, merchantability or fitness for any
-    particular purpose.
-
-    Please cite the author in any work or product based on this material.
-
-   ***************************************************************************
-
-    Modification History:
-           20 August of 1996 - Shavirin  -  originaly written
-
-    Bugs and restriction on use:
-
-    Notes:
-
-*****************************************************************************/
-#include <ncbi.h>
-
-#define INCLENGTH 256
-#define STRLENGTH 128
-
-typedef struct MDeflineID {
-  Uint4   gi;
-  CharPtr defline;
-  struct MDeflineID *next;
-} MDeflineID, PNTR MDeflineIDPtr;
-
-typedef struct FMergeID {
-  MDeflineIDPtr id;
-  CharPtr sequence;
-} FMergeID, PNTR FMergeIDPtr;
-
-Uint4Ptr GetIndexFromFasta(FILE *fd, Uint4Ptr MaxIndexCount);
-Uint4Ptr GetIndexFromIndex(FILE *fd, Uint4Ptr MaxIndexCount);
-
-static  FMergeIDPtr NextFastaFromFasta(FILE *fd);
-Boolean IfGiInIndex(Uint4Ptr index, Uint4 MaxIndexCount, Uint4 gi);
-Int4 MergeFMergeID(FMergeIDPtr fasta, 
-                   FILE *fd, FILE *fd_ind,
-                   Uint4Ptr index, 
-                   Uint4 MaxIndexCount);
-void FMergeIDFree(FMergeIDPtr fasta);
-
-#define NUMARG 5
-
-Args dump_args[NUMARG] = {
-  {"Job mode (this value must be set):\n"
-   "         1 - create index only\n"
-   "         2 - update database from existing index\n"
-   "         3 - create index and update database\n"
-   "                                   ",
-
-   NULL, NULL,NULL,FALSE,'t',ARG_STRING, 0.0,0,NULL},
-  {"Logfile name:","fmerge.log",
-     NULL,NULL,TRUE,'l',ARG_FILE_OUT, 0.0,0,NULL},
-  {"Input file for indexing\n"
-   "         Name of FASTA file, that you are going to update\n"
-   "                                   ",
-   "nr", NULL,NULL,TRUE,'n',ARG_FILE_IN,0.0,0,NULL},
-
-  {"Input file for new gis:\n"
-   "         Name of FASTA file with new sequences\n"
-   "                                   ", 
-   "month.aa", NULL,NULL,TRUE,'m',ARG_FILE_IN,0.0,0,NULL},
-  {"Index file name\n"
-   "         Name of file for index storage\n"
-   "                                   ",
-   "index.nr", NULL,NULL,TRUE,'i',ARG_FILE_IN,0.0,0,NULL},
-};
-
-#define JobModeStr  (const char *) dump_args[0].strvalue 
-#define LogFileName (const char *) dump_args[1].strvalue  
-#define NRFile      (const char *) dump_args[2].strvalue
-#define MonthFile   (const char *) dump_args[3].strvalue
-#define IndexFile   (const char *) dump_args[4].strvalue
-
-#define INDEX_MODE  1
-#define UPDATE_MODE 2
-#define FULL_MODE   3
-
-/* ------------------------------------------------------------------
-                This is handler for HeapSort function
-   ------------------------------------------------------------------*/
-static int LIBCALLBACK intcompare(VoidPtr i, VoidPtr j)
-{
-  if (*(int *)i > *(int *)j)
-    return (1);
-  if (*(int *)i < *(int *)j)
-    return (-1);
-  return (0);
-}
-
-Int2 Main(void) 
-{
-  Uint4Ptr index = NULL;
-  FILE *fd, *fd_ind, *fdout;
-  Uint4 i, MaxIndexCount;
-  FMergeIDPtr fasta;
-  Int4 Added =0, TotalAdded =0, SequenceAdded =0;
-  Int4 JobMode;
-
-  if ( !GetArgs ("fmerge",NUMARG,dump_args) ) {
-    return 1;
-  }
-  if ( !ErrSetLog (LogFileName) ) {
-    ErrShow();
-  } else {
-    ErrSetOpts (ERR_CONTINUE, ERR_LOG_ON);
-  }
-
-  JobMode = atoi(JobModeStr);
-
-  switch(JobMode) {
-  case INDEX_MODE:
-    ErrLogPrintf("fmerge started in \"Create index\" mode\n");
-    break;
-  case UPDATE_MODE:
-    ErrLogPrintf("fmerge started in \"Update database\" mode\n");
-    break;
-  case FULL_MODE:
-    ErrLogPrintf("fmerge started in \"Full mode\" mode\n");
-    break;
-  default:
-    printf("fmerge mode \"%s\" invalid, exiting...\n", 
-           JobModeStr);
-    exit(1);
-  }
-
-  /* First creating index if necessary */
-
-  if(JobMode == INDEX_MODE || JobMode == FULL_MODE) {
-
-    if((fd = FileOpen(NRFile, "r")) == NULL) {
-      ErrLogPrintf("Cannot open input file. Indexing failed...\n");
-      exit(1);
-    }
-    ErrLogPrintf("Using file \"%s\" to create index\n", NRFile);
-    
-    if((index = GetIndexFromFasta(fd, &MaxIndexCount)) == NULL) {
-      ErrLogPrintf("Error in creating gi index from FASTA file\n");
-    }
-    FileClose(fd);
-    ErrLogPrintf("Created index of %d gis\n", MaxIndexCount);
-    
-    fd_ind = FileOpen(IndexFile, "w");
-    for(i = 0; i < MaxIndexCount; i++)
-      fprintf(fd_ind, "%ld\n", index[i]); 
-    FileClose(fd_ind);
-	
-    if(JobMode == INDEX_MODE) {
-      MemFree(index);
-      exit(0);
-    }
-  }
-
-  if(JobMode == UPDATE_MODE) {
-    
-    if((fd_ind = FileOpen(IndexFile, "r")) == NULL) {
-      ErrLogPrintf("Cannot open file with index. Exiting...\n");
-      exit(1);
-    }
-    if((index = GetIndexFromIndex(fd_ind, &MaxIndexCount)) == NULL) {
-      ErrLogPrintf("Error in creating gi index from FASTA file\n");
-    }
-    FileClose(fd_ind);
-    ErrLogPrintf("Created index of %d gis\n", MaxIndexCount);
-  }
-
-  /* Now searching FASTA file with new gis*/
-
-  if((fd = FileOpen(MonthFile, "r")) == NULL) {
-    ErrLogPrintf("Cannot open input file. Indexing failed...\n");
-    exit(1);
-  }
-  ErrLogPrintf("Using file \"%s\" as FASTA database with new gis\n", 
-               MonthFile);
-
-  if((fdout = FileOpen(NRFile, "a+")) == NULL) {
-    ErrLogPrintf("Error opening %s file\n", NRFile);
-    exit(1);
-  }
-  
-  if((fd_ind = FileOpen(IndexFile, "a+")) == NULL) {
-    ErrLogPrintf("Error opening %s file\n", IndexFile);
-    exit(1);
-  }
-
-  while((fasta = NextFastaFromFasta(fd)) != NULL) {
-    if((Added = MergeFMergeID(fasta, fdout, fd_ind, 
-                              index, MaxIndexCount)) != 0) {
-      TotalAdded += Added;
-      SequenceAdded++;
-    }
-  }
-  MemFree(index);
-  FileClose(fd);
-  FileClose(fd_ind);
-  ErrLogPrintf("Total gis added:       %d\n", TotalAdded);
-  ErrLogPrintf("Total sequences added: %d\n", SequenceAdded);
-  return 0;
-}
-Uint4Ptr GetIndexFromIndex(FILE *fd, Uint4Ptr MaxIndexCount)
-{
- Uint4Ptr index;
- Uint4 IndexSize = INCLENGTH;
- Uint4 i =0, gi;
-
- index = MemNew(IndexSize*4);
- 
- while (fscanf(fd, "%ld", &gi) > 0) {
-   
-   if(i == IndexSize) {
-     IndexSize += INCLENGTH;
-     index = Realloc(index, IndexSize*4);
-   }
-   index[i] = gi;
-   i++;
- }
- *MaxIndexCount = i;
- HeapSort(index, *MaxIndexCount, sizeof(Uint4), intcompare);
- return index;
-}
-
-Uint4Ptr GetIndexFromFasta(FILE *fd, Uint4Ptr MaxIndexCount)
-{
-  Char str[20000];
-  CharPtr pos;
-  CharPtr start;
-  Uint4Ptr index;
-  Uint4 i =0, gi;
-  Uint4 IndexSize = INCLENGTH;
-
-  index = MemNew(IndexSize*4);
-
-  while(fgets (str, sizeof (str), fd) != NULL) {
-    start = str;
-    while((pos = StringStr(start, "\1gi|")) != NULL  ||
-          (pos = StringStr(start, ">gi|")) != NULL) {
-      if(!sscanf(pos+4, "%ld", &gi)) {
-        ErrLogPrintf("Error parsing gi number. Indexing failed...\n");
-        exit(1);
-      }
-      /* Here we got next gi nnumber and will add it to the index */
-
-      if(i == IndexSize) {
-        IndexSize += INCLENGTH;
-        index = Realloc(index, IndexSize*4);
-      }
-      index[i] = gi;
-      i++;
-      start = pos+4;
-    }
-  }
-  *MaxIndexCount = i;
-  HeapSort(index, *MaxIndexCount, sizeof(Uint4), intcompare);
-  return index;
-}
-
-Boolean IfGiInIndex(Uint4Ptr index, Uint4 MaxIndexCount, Uint4 gi)
-{
-  Uint4 high_index, low_index, new_index; 
-
-  if(MaxIndexCount == 0 || index == NULL || gi == 0)
-    return FALSE;
-
-  low_index  = 0;
-  high_index = MaxIndexCount;
-  new_index  = (low_index+high_index)/2;
-
-  while(new_index != low_index && new_index != high_index){
-    if (gi > index[new_index]) {
-      low_index = new_index;
-    } else if (gi < index[new_index]) {
-      high_index = new_index;
-    } else {	/* scores are equal. */
-      return TRUE;
-    }
-    new_index = (low_index+high_index)/2;
-  }
-  if (gi == index[new_index]) 
-    return TRUE;
-  else
-    return FALSE;
-}
-
-static FMergeIDPtr NextFastaFromFasta(FILE *fd)
-{
-  CharPtr str, start;
-  Int4 i=0,j=0;
-  Char ch;
-  Int4 gi;
-  CharPtr pos;
-  Int4 f_pos;
-  Char tmpbuff[2048];
-  FMergeIDPtr fasta;
-  MDeflineIDPtr id, last_id;
-  CharPtr TmpSequence;
-  Int4 SeqSize = STRLENGTH;
-
-  if((fasta = MemNew(sizeof(FMergeID))) == NULL) {
-    ErrLogPrintf("Cannot allocate memory for Fasta ID\n");
-    exit(1);
-  }
-  fasta->id = NULL;
-
-  str = MemNew(SeqSize + 5);
-  str[0] = NULLB;
-  for(i=0; (FileRead(&ch, sizeof(ch), 1, fd) == 1); i++) {      
-    if((str[i] = ch) == '\n' || ch == '\r')
-      break;
-    if (i > SeqSize) {
-      SeqSize = i + STRLENGTH;
-      str = Realloc(str, SeqSize + 5);
-    }
-  }
-  str[i+1] = NULLB;
-  start = str;
-  while((pos = StringStr(str, "\1gi|")) != NULL ||
-        (pos = StringStr(str, ">gi|")) != NULL) {
-    if(!sscanf(pos+4, "%ld", &gi)) {
-      ErrLogPrintf("Error parsing gi number. Indexing failed...\n");
-      exit(1);
-    }
-    pos++;
-    for(j = 0;pos[j] != '\1' && 
-              pos[j] != '\n' && 
-              pos[j] != '\r'; j++) {
-      tmpbuff[j] = pos[j];
-    }
-    tmpbuff[j] = '\0';
-
-    id = MemNew(sizeof(MDeflineID));
-    id->defline = StringSave(tmpbuff);
-    id->gi = gi;
-    id->next = NULL;
-
-    if(fasta->id == NULL) {
-      fasta->id = id;
-      last_id = fasta->id;
-    } else {
-      last_id->next = id;
-      last_id = last_id->next;
-    }
-    str = pos+4;
-  }
-
-  MemFree(start);
-
-  if(fasta->id == NULL) {
-    MemFree(fasta);
-    return NULL;
-  }
-
-  TmpSequence = MemNew(SeqSize + 5);
-  TmpSequence[0] = NULLB;
-
-
-  while(TRUE) {
-    f_pos = ftell(fd);
-    if(fgets(tmpbuff, sizeof (tmpbuff), fd) == NULL)
-      break;
-    if(StringStr(tmpbuff, ">gi|") != NULL) { /* new fasta entry started */
-      fseek(fd, (long) f_pos, SEEK_SET);
-      break;
-    }
-    StringCat(TmpSequence, tmpbuff);
-    SeqSize += STRLENGTH + 5;
-    TmpSequence = Realloc(TmpSequence, SeqSize);
-  }
-  fasta->sequence = TmpSequence;
-  return fasta;
-}
-
-Int4 MergeFMergeID(FMergeIDPtr fasta, FILE *fd, FILE *fd_ind,
-                   Uint4Ptr index, Uint4 MaxIndexCount) 
-{
-  Boolean SomeNew = FALSE;
-  MDeflineIDPtr id;
-  Int4 TotalNot =0;
-
-  id = fasta->id;
-  while(id != NULL) {
-    if(IfGiInIndex(index, MaxIndexCount, id->gi)) {
-      id = id->next;
-      continue;
-    }
-    TotalNot++;
-
-    fprintf(fd_ind, "%ld\n", id->gi);
-
-    if(SomeNew)
-      fprintf(fd, "\1%s", id->defline);
-    else {
-      fprintf(fd, ">%s", id->defline);
-      SomeNew = TRUE;
-    }
-    id = id->next;
-  }
-
-  if(SomeNew)
-    fprintf(fd, "\n%s", fasta->sequence);
-  
-  FMergeIDFree(fasta);
-  return TotalNot;
-}
-void FMergeIDFree(FMergeIDPtr fasta)
-{
-  MDeflineIDPtr id, id1;
-
-  MemFree(fasta->sequence);
-  id = fasta->id;
-  while(id != NULL) {
-    MemFree(id->defline);
-    id1 = id;
-    id = id->next;
-    MemFree(id1);
-  }
-  MemFree(fasta);
-}
-
diff --git a/demo/makemat.c b/demo/makemat.c
index dba93c25..49939238 100644
--- a/demo/makemat.c
+++ b/demo/makemat.c
@@ -1,4 +1,4 @@
-static char const rcsid[] = "$Id: makemat.c,v 6.13 2003/05/30 17:31:09 coulouri Exp $";
+static char const rcsid[] = "$Id: makemat.c,v 6.14 2004/05/14 12:10:06 camacho Exp $";
 
 /*
 * ===========================================================================
@@ -180,6 +180,10 @@ static Boolean takeMatrixCheckpoint(compactSearchItems * compactSearch,
         localChar = getRes(compactSearch->query[i]);
 
         fprintf(checkFile,"%c",localChar);
+
+        /* The following 2 lines are needed to preserve compatibility with the
+         * checkpoint file libraries distributed with IMPALA (from personal
+         * communication with IMPALA's author) */
         posSearch->posMatrix[i][Xchar] = Xscore;
         posSearch->posPrivateMatrix[i][Xchar] = Xscore * scalingFactor;
     }  
@@ -322,6 +326,8 @@ Nlm_FloatHi scalingFactor, Char *directoryPrefix)
         BlastScoreBlkMatFill(sbp, underlyingMatrixName);
         compactSearch->matrix = sbp->matrix;
         compactSearch->gapped_calculation = TRUE;
+        /* Note that these two assignments are not really needed for
+         * makemat's operation and thus their values are irrelevant */
         compactSearch->pseudoCountConst = 10;
         compactSearch->ethresh = 0.001;
         BlastScoreBlkFill(sbp,  (CharPtr) query, queryLength, 0);
diff --git a/demo/megablast.c b/demo/megablast.c
index 06778dcc..d141a337 100644
--- a/demo/megablast.c
+++ b/demo/megablast.c
@@ -1,6 +1,6 @@
-static char const rcsid[] = "$Id: megablast.c,v 6.113 2004/04/29 19:56:00 dondosha Exp $";
+static char const rcsid[] = "$Id: megablast.c,v 6.114 2004/05/27 17:37:30 dondosha Exp $";
 
-/* $Id: megablast.c,v 6.113 2004/04/29 19:56:00 dondosha Exp $
+/* $Id: megablast.c,v 6.114 2004/05/27 17:37:30 dondosha Exp $
 **************************************************************************
 *                                                                         *
 *                             COPYRIGHT NOTICE                            *
@@ -28,6 +28,9 @@ static char const rcsid[] = "$Id: megablast.c,v 6.113 2004/04/29 19:56:00 dondos
 ************************************************************************** 
  * $Revision 6.13$ *  
  * $Log: megablast.c,v $
+ * Revision 6.114  2004/05/27 17:37:30  dondosha
+ * Do not call GapXEditBlockDelete in formatting callback - this is now done when HSPs are freed
+ *
  * Revision 6.113  2004/04/29 19:56:00  dondosha
  * Mask filtered locations in query sequence lines in XML output
  *
@@ -653,7 +656,6 @@ MegaBlastPrintSegments(VoidPtr ptr)
       hsp = search->current_hitlist->hsp_array[hsp_index];
       if (hsp==NULL || (search->pbp->cutoff_e > 0 && 
                         hsp->evalue > search->pbp->cutoff_e)) {
-         GapXEditBlockDelete(hsp->gap_info); /* Don't need it anymore */
 	 continue;
       }
       context = hsp->context;
@@ -748,7 +750,6 @@ MegaBlastPrintSegments(VoidPtr ptr)
       GXECollectDataForSeqalign(hsp->gap_info, hsp->gap_info->esp, numseg,
 				&start, &length, &strands, 
 				&q_off, &hsp->subject.offset);
-      GapXEditBlockDelete(hsp->gap_info); /* Don't need it anymore */
 
       if (start[0] < 0) {
          length[0] += start[0];
diff --git a/demo/tbl2asn.c b/demo/tbl2asn.c
index b95dc25c..3ea959d5 100644
--- a/demo/tbl2asn.c
+++ b/demo/tbl2asn.c
@@ -29,7 +29,7 @@
 *
 * Version Creation Date:   5/5/00
 *
-* $Revision: 6.85 $
+* $Revision: 6.88 $
 *
 * File Description: 
 *
@@ -1428,12 +1428,14 @@ static Uint2 ProcessDeltaSet (
   Uint2          entityID;
   SeqEntryPtr    firstsep, lastsep, nextsep, sep, topsep;
   IntFuzzPtr     ifp;
+  Boolean        is_unk100;
+  ObjectIdPtr    oip;
   ObjMgrDataPtr  omdptop;
   ObjMgrData     omdata;
   Uint2          parenttype;
   Pointer        parentptr;
   CharPtr        seqbuf;
-  SeqIdPtr       sip;
+  SeqIdPtr       sip, virtid;
   SeqLitPtr      slp;
   ValNodePtr     vnp;
 
@@ -1491,7 +1493,7 @@ static Uint2 ProcessDeltaSet (
   bsp = FindNucBioseq (firstsep);
   if (bsp == NULL) return 0;
 
-  sip = SeqIdDup (bsp->id);
+  sip = SeqIdSetDup (bsp->id);
   vnp = ValNodeExtract (&(bsp->descr), Seq_descr_title);
 
   deltabsp = BioseqNew ();
@@ -1554,8 +1556,21 @@ static Uint2 ProcessDeltaSet (
 
       slp->length = bsp->length;
       ValNodeAddPointer ((ValNodePtr PNTR) &(deltabsp->seq_ext), (Int2) 2, (Pointer) slp);
-      if (slp->length < 1) {
-        slp->length = 0;
+
+      is_unk100 = FALSE;
+      virtid = bsp->id;
+      if (virtid != NULL && virtid->choice == SEQID_LOCAL) {
+        oip = (ObjectIdPtr) virtid->data.ptrvalue;
+        if (oip != NULL) {
+          if (StringCmp (oip->str, "unk100") == 0) {
+            is_unk100 = TRUE;
+          }
+        }
+      }
+      if (slp->length < 1 || is_unk100) {
+        if (slp->length < 1) {
+          slp->length = 0;
+        }
         ifp = IntFuzzNew ();
         ifp->choice = 4;
         slp->fuzz = ifp;
@@ -3126,6 +3141,7 @@ Int2 Main (void)
 
 {
   AsnIoPtr        aip = NULL;
+  Char            app [64];
   CharPtr         base;
   AsnTypePtr      bssp_atp = NULL;
   CitSubPtr       csp;
@@ -3185,7 +3201,8 @@ Int2 Main (void)
 
   /* process command line arguments */
 
-  if (! GetArgs ("tbl2asn", sizeof (myargs) / sizeof (Args), myargs)) {
+  sprintf (app, "tbl2asn %s", TBL2ASN_APPLICATION);
+  if (! GetArgs (app, sizeof (myargs) / sizeof (Args), myargs)) {
     return 0;
   }