Changed default reflen; more informative error message when GCD bin size too small...

[samtools.git] / misc / bamcheck.c

diff --git a/misc/bamcheck.c b/misc/bamcheck.c
index 66a6861716a6f3aa1be208acae460530090d2b96..a438adac3453e73a86f65017bb72ca54ba3ac4c4 100644 (file)
--- a/misc/bamcheck.c
+++ b/misc/bamcheck.c
@@ -293,7 +293,7 @@ void count_indels(stats_t *stats,bam1_t *bam_line)
 
         if ( cig==1 )
         {
-            int idx = is_fwd ? icycle : read_len-icycle;
+            int idx = is_fwd ? icycle : read_len-icycle-ncig;
             if ( idx<0 ) 
                 error("FIXME: read_len=%d vs icycle=%d\n", read_len,icycle);
             if ( idx >= stats->nbases || idx<0 ) error("FIXME: %d vs %d, %s:%d %s\n", idx,stats->nbases, stats->sam->header->target_name[bam_line->core.tid],bam_line->core.pos+1,bam1_qname(bam_line));
@@ -515,9 +515,9 @@ void realloc_gcd_buffer(stats_t *stats, int seq_len)
 
     int n = 1 + stats->gcd_ref_size / (stats->gcd_bin_size - seq_len);
     if ( n <= stats->igcd )
-        error("Uh: n=%d igcd=%d\n", n,stats->igcd );
+        error("The --GC-depth bin size is too small or reference genome too big; please decrease the bin size or increase the reference length\n");
         
-    if ( n >= stats->ngcd )
+    if ( n > stats->ngcd )
     {
         stats->gcd = realloc(stats->gcd, n*sizeof(gc_depth_t));
         if ( !stats->gcd )
@@ -1043,6 +1043,8 @@ void output_stats(stats_t *stats)
     printf("# Indels per cycle. Use `grep ^IC | cut -f 2-` to extract this part. The columns are: cycle, number of insertions (fwd), .. (rev) , number of deletions (fwd), .. (rev)\n");
     for (ilen=0; ilen<=stats->nbases; ilen++)
     {
+        // For deletions we print the index of the cycle before the deleted base (1-based) and for insertions
+        //  the index of the cycle of the first inserted base (also 1-based)
         if ( stats->ins_cycles_1st[ilen]>0 || stats->ins_cycles_2nd[ilen]>0 || stats->del_cycles_1st[ilen]>0 || stats->del_cycles_2nd[ilen]>0 )
             printf("IC\t%d\t%ld\t%ld\t%ld\t%ld\n", ilen+1, (long)stats->ins_cycles_1st[ilen], (long)stats->ins_cycles_2nd[ilen], (long)stats->del_cycles_1st[ilen], (long)stats->del_cycles_2nd[ilen]);
     }
@@ -1277,7 +1279,7 @@ void error(const char *format, ...)
         printf("    -d, --remove-dups                   Exlude from statistics reads marked as duplicates\n");
         printf("    -f, --required-flag <int>           Required flag, 0 for unset [0]\n");
         printf("    -F, --filtering-flag <int>          Filtering flag, 0 for unset [0]\n");
-        printf("        --GC-depth <float,float>        Bin size for GC-depth graph and the maximum reference length [2e4,6e9]\n");
+        printf("        --GC-depth <float,float>        Bin size for GC-depth graph and the maximum reference length [2e4,4.2e9]\n");
         printf("    -h, --help                          This help message\n");
         printf("    -i, --insert-size <int>             Maximum insert size [8000]\n");
         printf("    -I, --id <string>                   Include only listed read group or sample name\n");
@@ -1316,7 +1318,7 @@ int main(int argc, char *argv[])
     stats->max_qual  = 40;
     stats->isize_main_bulk = 0.99;   // There are always outliers at the far end
     stats->gcd_bin_size = 20e3;
-    stats->gcd_ref_size = 3e9;
+    stats->gcd_ref_size = 4.2e9;
     stats->rseq_pos     = -1;
     stats->tid = stats->gcd_pos = stats->igcd = -1;
     stats->is_sorted = 1;