Merge pull request #128 from andrewjpage/update_robustness

Update robustness

VERSION

@@ -1 +1 @@
-1.1.1
+1.1.2

debian/changelog

@@ -1,8 +1,14 @@
+gubbins (1.1.2~trusty1) trusty; urgency=low
+
+ * Fix GFF formatting and C warnings
+
+ -- Andrew Page <ap13@sanger.ac.uk> Fri, 23 Jan 2015 16:02:00 +0000
+
 gubbins (1.1.1~trusty1) trusty; urgency=low
 
  * Message to tell user theres a problem with outgroups not a clade
 
- -- Andrew Page <ap13@sanger.ac.uk> Fri, 23 Jan 2015 16:02 gubbins (1.1.0~trusty1) trusty; urgency=low
+ -- Andrew Page <ap13@sanger.ac.uk> Fri, 23 Jan 2015 16:02:00 +0000
 
 gubbins (1.1.0~trusty1) trusty; urgency=low
 

install-userspace.sh

@@ -13,7 +13,7 @@
 #
 
 py_pkgs=( "biopython" "dendropy" )
-deb_urls=( "http://uk.archive.ubuntu.com/ubuntu/pool/universe/r/raxml/raxml_7.2.8-2_amd64.deb" "https://launchpad.net/~ap13/+archive/ubuntu/gubbins/+files/fastml2_2.2~trusty1_amd64.deb" "https://launchpad.net/~ap13/+archive/ubuntu/gubbins/+files/gubbins_0.1.7~trusty2_amd64.deb" )
+deb_urls=( "http://uk.archive.ubuntu.com/ubuntu/pool/universe/r/raxml/raxml_7.2.8-2_amd64.deb" "https://launchpad.net/~ap13/+archive/ubuntu/gubbins/+files/fastml2_2.3~trusty1_amd64.deb" "https://launchpad.net/~ap13/+archive/ubuntu/gubbins/+files/gubbins_1.1.1~trusty1_amd64.deb" )
 
 function check_platform {
  # Ubuntu 14.04

python/gubbins/common.py

@@ -17,24 +17,23 @@
 # Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
 #
 
-import sys
-import argparse
-import subprocess
-import os
-import time
-import re
-import tempfile
-from collections import Counter
+from Bio import AlignIO
 from Bio import Phylo
-import dendropy
 from Bio import SeqIO
-from Bio import AlignIO
 from Bio.Align import MultipleSeqAlignment
 from Bio.Seq import Seq
 from cStringIO import StringIO
+from collections import Counter
+import argparse
+import dendropy
+import math
+import os
+import re
 import shutil
 import subprocess
-import math
+import sys
+import tempfile
+import time
 
 class GubbinsError(Exception):
  def __init__(self, value,message):
@@ -205,7 +204,11 @@ def parse_and_run(self):
  # find all snp sites
  if self.args.verbose > 0:
  print GUBBINS_EXEC +" "+ self.args.alignment_filename
- subprocess.check_call([GUBBINS_EXEC, self.args.alignment_filename])
+ try:
+ subprocess.check_call([GUBBINS_EXEC, self.args.alignment_filename])
+ except:
+ sys.exit("Gubbins crashed, please ensure you have enough free memory")
+
  if self.args.verbose > 0:
  print int(time.time())
 
@@ -282,7 +285,10 @@ def parse_and_run(self):
  if self.args.starting_tree is not None and i == 1:
  shutil.copyfile(self.args.starting_tree, current_tree_name)
  else:
- subprocess.check_call(tree_building_command, shell=True)
+ try:
+ subprocess.check_call(tree_building_command, shell=True)
+ except:
+ sys.exit("Failed while building the tree.")
 
  if self.args.verbose > 0:
  print int(time.time())
@@ -295,7 +301,11 @@ def parse_and_run(self):
  fastml_command_suffix = ''
 
 
- subprocess.check_call(fastml_command+fastml_command_suffix, shell=True)
+ try:
+ subprocess.check_call(fastml_command+fastml_command_suffix, shell=True)
+ except:
+ sys.exit("Failed while running FastML")
+
  shutil.copyfile(current_tree_name+'.output_tree',current_tree_name)
  shutil.copyfile(starting_base_filename+".start", starting_base_filename+".gaps.snp_sites.aln")
  GubbinsCommon.reinsert_gaps_into_fasta_file(current_tree_name+'.seq.joint.txt', starting_base_filename +".gaps.vcf", starting_base_filename+".gaps.snp_sites.aln")
@@ -309,7 +319,10 @@ def parse_and_run(self):
 
  if self.args.verbose > 0:
  print gubbins_command
- subprocess.check_call(gubbins_command, shell=True)
+ try:
+ subprocess.check_call(gubbins_command, shell=True)
+ except:
+ sys.exit("Failed while running Gubbins. Please ensure you have enough free memory")
  if self.args.verbose > 0:
  print int(time.time())
 
@@ -941,10 +954,16 @@ def pairwise_comparison(filename,base_filename,gubbins_exec,alignment_filename,f
  sequence_names = GubbinsCommon.get_sequence_names_from_alignment(filename)
  GubbinsCommon.create_pairwise_newick_tree(sequence_names, base_filename+".tre")
 
- subprocess.check_call(GubbinsCommon.generate_fastml_command(fastml_exec, base_filename+".gaps.snp_sites.aln", base_filename+".tre"), shell=True)
+ try:
+ subprocess.check_call(GubbinsCommon.generate_fastml_command(fastml_exec, base_filename+".gaps.snp_sites.aln", base_filename+".tre"), shell=True)
+ except:
+ sys.exit("Failed while running fastML")
  shutil.copyfile(base_filename+'.tre.output_tree',base_filename+".tre")
  shutil.copyfile(base_filename+'.tre.seq.joint.txt', base_filename+".snp_sites.aln")
- subprocess.check_call(gubbins_exec+" -r -v "+base_filename+".vcf -t "+base_filename+".tre -f "+ alignment_filename +" "+ base_filename+".snp_sites.aln", shell=True)
+ try:
+ subprocess.check_call(gubbins_exec+" -r -v "+base_filename+".vcf -t "+base_filename+".tre -f "+ alignment_filename +" "+ base_filename+".snp_sites.aln", shell=True)
+ except:
+ sys.exit("Failed while running Gubbins")
  GubbinsCommon.rename_files(GubbinsCommon.translation_of_filenames_to_final_filenames_pairwise(base_filename, base_filename_without_ext))
 
  @staticmethod

python/gubbins/tests/data/expected_RAxML_result.multiple_recombinations.iteration_5.gff

@@ -1,6 +1,6 @@
 ##gff-version 3
 ##sequence-region SEQUENCE 1 242
-SEQUENCE GUBBINS CDS 29 49 0.000 . 0 node="N7->sequence_6";neg_log_likelihood="4.955311"taxa="sequence_6";snp_count="21"
-SEQUENCE GUBBINS CDS 29 84 0.000 . 0 node="N5->N6";neg_log_likelihood="12.082148"taxa=" sequence_9 sequence_6 sequence_8 sequence_7";snp_count="51"
-SEQUENCE GUBBINS CDS 51 84 0.000 . 0 node="N4->N5";neg_log_likelihood="10.195830"taxa=" sequence_5 sequence_9 sequence_6 sequence_8 sequence_7";snp_count="30"
-SEQUENCE GUBBINS CDS 51 84 0.000 . 0 node="N1->N4";neg_log_likelihood="10.195830"taxa=" sequence_1 sequence_5 sequence_9 sequence_6 sequence_8 sequence_7";snp_count="30"
+SEQUENCE GUBBINS CDS 29 49 0.000 . 0 node="N7->sequence_6";neg_log_likelihood="4.955311";taxa="sequence_6";snp_count="21";
+SEQUENCE GUBBINS CDS 29 84 0.000 . 0 node="N5->N6";neg_log_likelihood="12.082148";taxa=" sequence_9 sequence_6 sequence_8 sequence_7";snp_count="51";
+SEQUENCE GUBBINS CDS 51 84 0.000 . 0 node="N4->N5";neg_log_likelihood="10.195830";taxa=" sequence_5 sequence_9 sequence_6 sequence_8 sequence_7";snp_count="30";
+SEQUENCE GUBBINS CDS 51 84 0.000 . 0 node="N1->N4";neg_log_likelihood="10.195830";taxa=" sequence_1 sequence_5 sequence_9 sequence_6 sequence_8 sequence_7";snp_count="30";

release/manifests/trustyvm.pp

@@ -2,7 +2,7 @@
  ensure => "installed"
  }
 
-package { ["gcc", "build-essential", "pkg-config"]:
+package { ["gcc", "build-essential", "pkg-config","ntp"]:
  ensure => "installed"
  }
 

src/alignment_file.c

@@ -28,6 +28,7 @@
 #include "vcf.h"
 #include "alignment_file.h"
 #include "snp_sites.h"
+#include "string_cat.h"
 
 KSEQ_INIT(gzFile, gzread)
 

src/gff_file.c

@@ -37,9 +37,9 @@ void print_gff_line(FILE * gff_file_pointer, int start_coordinate, int end_coord
  fprintf(gff_file_pointer, "0.000\t.\t0\t");
 
  fprintf(gff_file_pointer, "node=\"%s->%s\";", parent_node_id, current_node_id );
- fprintf(gff_file_pointer, "neg_log_likelihood=\"%f\"", neg_log_likelihood);
+ fprintf(gff_file_pointer, "neg_log_likelihood=\"%f\";", neg_log_likelihood);
  fprintf(gff_file_pointer, "taxa=\"%s\";", taxon_names);
- fprintf(gff_file_pointer, "snp_count=\"%d\"", number_of_snps);
+ fprintf(gff_file_pointer, "snp_count=\"%d\";", number_of_snps);
  fprintf(gff_file_pointer, "\n");
 
  fflush(gff_file_pointer);

src/gubbins.c

@@ -31,7 +31,8 @@
 #include "tree_scaling.h"
 #include "seqUtil.h"
 #include "Newickform.h"
-
+#include "tree_statistics.h"
+#include "fasta_of_snp_sites.h"
 
 
 // get reference sequence from VCF, and store snp locations

src/main.c

@@ -49,11 +49,6 @@ void print_usage(FILE* stream, int exit_code)
  " -m Min SNPs for identifying a recombination block\n"
  " -h Display this usage information.\n\n"
 );
-
- fprintf (stream, "Step 1: Detect SNP sites (generates inputs files for step 2)\n");
- fprintf (stream, "gubbins alignment_file\n\n", program_name);
- fprintf (stream, "Step 2: Detect recombinations\n");
- fprintf (stream, "gubbins -r -v vcf_file -t newick_tree -f original.aln -m 10 alignment_file\n\n", program_name);
  exit (exit_code);
 }
 
@@ -64,6 +59,7 @@ int check_file_exists_or_exit(char * filename)
  } else {
  printf("Error: File '%s' doesnt exist\n",filename);
  print_usage(stderr, EXIT_FAILURE);
+ return 0;
  }
 }
 

src/parse_phylip.h

@@ -50,6 +50,8 @@ int get_internal_node(int sequence_index);
 void fill_in_unambiguous_bases_in_parent_from_children_where_parent_has_a_gap(int parent_sequence_index, int * child_sequence_indices, int num_children);
 void fill_in_unambiguous_gaps_in_parent_from_children(int parent_sequence_index, int * child_sequence_indices, int num_children);
 void freeup_memory();
+void set_number_of_bases_in_recombinations(char * sample_name, int bases_in_recombinations);
+void filter_sequence_bases_and_rotate(char * reference_bases, char ** filtered_bases_for_snps, int number_of_filtered_snps);
 
 #define MAX_READ_BUFFER 65536
 #define MAX_SAMPLE_NAME_SIZE 1024

src/seqUtil.c

@@ -1,6 +1,7 @@
 #define __SEQUTIL_C__
 
 #include "seqUtil.h"
+#include "string_cat.h"
 
 /*
  * Yu-Wei Wu http://yuweibioinfo.blogspot.com/2008/10/newick-tree-parser-in-c-make-use-of.html

src/seqUtil.h

@@ -4,6 +4,7 @@
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
+#include "string_cat.h"
 
 /*
  * Yu-Wei Wu http://yuweibioinfo.blogspot.com/2008/10/newick-tree-parser-in-c-make-use-of.html

src/snp_searching.c

@@ -20,6 +20,7 @@
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
+#include <ctype.h>
 #include "snp_searching.h"
 
 // Most of the methods in this file look the same, so should be DRYed out.

src/snp_searching.h

@@ -33,5 +33,7 @@ int advance_window_start_to_next_snp_with_start_index(int window_start_coordinat
 int rewind_window_end_to_last_snp_with_start_end_index(int window_end_coordinate, int * snp_locations, char * child_sequence, int number_of_branch_snps, int start_index,int end_index);
 int find_number_of_snps_in_block_with_start_end_index(int window_start_coordinate, int window_end_coordinate, int * snp_locations, char * child_sequence, int number_of_snps, int start_index,int end_index);
 int get_window_end_coordinates_excluding_gaps_with_start_end_index(int window_start_coordinate, int window_size, int * snp_locations, char * child_sequence, int number_of_snps, int start_index,int end_index);
+int calculate_block_size_without_gaps_with_start_end_index(char * child_sequence, int * snp_locations, int starting_coordinate, int ending_coordinate, int length_of_original_genome, int start_index,int end_index);
+
 
 #endif

src/snp_sites.c

@@ -28,6 +28,7 @@
 #include "phylip_of_snp_sites.h"
 #include "parse_phylip.h"
 #include "string_cat.h"
+#include "fasta_of_snp_sites.h"
 
 
 void build_snp_locations(int snp_locations[], char reference_sequence[])

src/vcf.c

@@ -168,7 +168,7 @@ void output_vcf_row_samples_bases(FILE * vcf_file_pointer, char reference_base,
  {
  continue;
  }
- if((bases_for_snp[i] == reference_base))
+ if(bases_for_snp[i] == reference_base)
  {
  fprintf( vcf_file_pointer, "%c", (char) reference_base ); 
  }

tests/data/one_recombination.tre.expected.gff

@@ -1,3 +1,3 @@
 ##gff-version 3
 ##sequence-region SEQUENCE 1 135
-SEQUENCE GUBBINS CDS 63 167 0.000 . 0 node="N5->sequence_2";taxa="sequence_2";snp_count="103"
+SEQUENCE GUBBINS CDS 63 167 0.000 . 0 node="N5->sequence_2";taxa="sequence_2";snp_count="103";