Changeset 8351 for trunk

Timestamp:

02/02/12 01:57:23 (4 months ago)

Author:

epruesse

Message:

keep a bit more data when cleaning an ARB DB for PT-server use
* the SAIs HELIX and HELIX_NR in addition to ECOLI
* all of presets/

We need the latter to keep the ARB file sufficiently complete to be opened
by arb_ntree. Previously, GBT_get_alignment_width would fail. NT_repair_DB
would cause a segmentation fault. Keeping all presets costs a bit of
memory/space, but the ability to open a database from lib/pts for
inspection seems more important than saving a few bytes (and fixing
NT_repair_DB much more difficult).

For the same reason, I think HELIX and HELIX_NR should be kept, as they
are expected by BI_helix. Since the cost is negligible, I'd argue for
keeping them just in case.

(Alternative solution, not pursued: always keep a cleaned and a full copy

of the ARB database in lib/pts, possibly changing the suffix on the
cleaned databases to express that they are no longer compatible)

Files:

• trunk/SL/PTCLEAN/ptclean.cxx (modified) (3 diffs)

trunk/SL/PTCLEAN/ptclean.cxx

@@ -46 +46 @@
                 if (strcmp(keyname, "species_data")  == 0) return SOME_OF_SPECIES_DATA;
                 if (strcmp(keyname, "extended_data") == 0) return SOME_OF_EXTENDED_DATA;
-                if (strcmp(keyname, "presets")       == 0) return SOME_OF_PRESETS;
+                if (strcmp(keyname, "presets")       == 0) return ALL;
                 break;
 
@@ -56 +56 @@
                 if (strcmp(keyname, "extended") == 0) {
                     const char *sainame = GBT_read_name(gbd);
-                    if (strcmp(sainame, "ECOLI") == 0) return SOME_OF_EXTENDED; // only "need" ECOLI 
+                    if (strcmp(sainame, "ECOLI") == 0) return SOME_OF_EXTENDED; 
+                    if (strcmp(sainame, "HELIX") == 0) return SOME_OF_EXTENDED;
+                    if (strcmp(sainame, "HELIX_NR") == 0) return SOME_OF_EXTENDED;
                 }
                 break;
@@ -69 +71 @@
                 if (strcmp(keyname, "full_name") == 0) return ALL;
                 if (strcmp(keyname, ali_name)    == 0) return SOME_OF_ALI_CONTAINER;
-                break;
-
-            case SOME_OF_PRESETS:
-                if (strcmp(keyname, "use") == 0) return ALL;
                 break;