pipeline: add NKMERS= to vcfcov

Fixes #45. Thanks to @peterdfields.

- ctx_vcfcov.c: ignore -n,--nkmers if --low-mem is passed
- update htslib
- add some comments to tests

scripts/make-pipeline.pl

@@ -375,16 +375,16 @@ sub print_usage
 STAMPY_BASE='.(defined($stampy_base) ? $stampy_base : '').'
 
 # Set up memory, threads and number of kmers in the graph
-CTX_ARGS=
 ifdef MEM
-  CTX_ARGS:=$(CTX_ARGS) -m $(MEM)
+  CTX_ARGS_MEM=-m $(MEM)
 endif
 ifdef NKMERS
-  CTX_ARGS:=$(CTX_ARGS) -n $(NKMERS)
+  CTX_ARGS_NKMERS=-n $(NKMERS)
 endif
 ifdef NTHREADS
-  CTX_ARGS:=$(CTX_ARGS) -t $(NTHREADS)
+  CTX_ARGS_THREADS=-t $(NTHREADS)
 endif
+CTX_ARGS=$(CTX_ARGS_MEM) $(CTX_ARGS_NKMERS) $(CTX_ARGS_THREADS)
 
 #
 # Parse USE_LINKS and JOINT_CALLING Makefile options
@@ -939,7 +939,7 @@ sub merge_vcf_list
         for my $assem (qw(links plain)) {
           my $callroot = "$call.$pop.$assem.$kmerstr";
           print "$proj/k$k/vcfcov/$callroot.%.vcf.gz: $proj/vcfs/$callroot.vcf.gz $proj/k$k/graphs/%.raw.ctx | dirs\n";
-          print "\t( $mccortex vcfcov -m \$(MEM) \$(VCFCOV_ARGS) --ref $ref_path --out-fmt vcfgz \$^ | \$(BCFTOOLS) view --samples \"\$*\" -o \$@ -O z ) >& \$@.log\n\n";
+          print "\t( $mccortex vcfcov \$(CTX_ARGS_MEM) \$(CTX_ARGS_NKMERS) \$(VCFCOV_ARGS) --ref $ref_path --out-fmt vcfgz \$^ | \$(BCFTOOLS) view --samples \"\$*\" -o \$@ -O z ) >& \$@.log\n\n";
         }
       }
     }

scripts/mccortex

@@ -4,8 +4,8 @@
 set -euo pipefail
 
 function usage {
-  echo "usage mccortex <K> ..." >&2
-  echo "  wrapper to find the correct mccortex binary given kmer size (K)" >&2
+  echo "usage: mccortex <K> [<cmd> ...]" >&2
+  echo "  Wrapper to find the correct mccortex binary given kmer size (K)" >&2
   exit -1
 }
 

src/commands/ctx_vcfcov.c

@@ -104,6 +104,13 @@ int ctx_vcfcov(int argc, char **argv)
   if(use_lowmem && use_himem)
     cmd_print_usage("Cannot use --low-mem and --high-mem together!");
 
+  // Override number of kmers to use if --low-mem passed, since we calculate
+  // number of kmers required anyway
+  if(use_lowmem && memargs.num_kmers_set) {
+    memargs.num_kmers_set = 0;
+    memargs.num_kmers_set = false;
+  }
+
   if(!max_allele_len) max_allele_len = DEFAULT_MAX_ALLELE_LEN;
   if(!max_gt_vars) max_gt_vars = DEFAULT_MAX_GT_VARS;
 

tests/pipeline/Makefile

@@ -1,5 +1,10 @@
 SHELL=/bin/bash -euo pipefail
 
+#
+# Simulate reads from two given samples, with sequencing error of 1%, 50X, 100bp
+# Call variants and check that we get exactly the expected variants
+#
+
 CTXDIR=../..
 MCCORTEX=$(CTXDIR)/bin/mccortex31
 DNACAT=$(CTXDIR)/libs/seq_file/bin/dnacat