transrate 0.3.1 → 1.0.0.alpha.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: be3eaa6170b268789a0ff6d6cb68448a7af6e037
-  data.tar.gz: 5e857b0392f9791e7111069ee1ed61ee19d3853c
+  metadata.gz: 329f398d7dc832c5e56dfe90b6483fb00321bfce
+  data.tar.gz: 52acb0232a8f42cb3604c2981899f0d931a1195b
 SHA512:
-  metadata.gz: f87811816aa975bfc85cfd025a90029c310bfdad19587cb5ea6a4909a78151f1904db0d69e5593b56f0f6b0044f5072e1e966580d7b6a45f78522ede8eda52cd
-  data.tar.gz: e9dba5e8a9910716250c3bd82bea190264d4ac32caf67acc44009b080e5eab553cf2a7317d1153d851f797c0eb45d07e1531896743b6bb1b3df902608d696d15
+  metadata.gz: 35f079c3474e02aa98a896cf6bbe4cd3d87ef7b7e95c91d13abea585eeb54b598ffe9539f7fc7b8c87a40ec78df92a8e1ba9fd23ac46e4da9c8f616919ab004e
+  data.tar.gz: fca938cc86122088ecb0308273d8552e1ebb26dd47e52e8154015aa27e68814016cedda45553f4bf76fcc9b164bcaaaf1cad1787c4f8296f316e6833b235e6bc

data/.gitignore

@@ -18,6 +18,7 @@ tmp
 .#*
 \#*
 *so
+dryrun
 
 # YARD artifacts
 .yardoc
@@ -39,3 +40,9 @@ dryrun
 *.bam
 *.csv
 *.coverage
+
+# c extension build artefacts
+Makefile
+transrate.bundle
+transrate.o
+.RUBYARCHDIR.time

data/README.md

@@ -4,11 +4,12 @@
 
 ## Development status
 
-[![Gem Version](https://badge.fury.io/rb/transrate.png)][gem]
-[![Build Status](https://secure.travis-ci.org/Blahah/transrate.png?branch=master)][travis]
-[![Dependency Status](https://gemnasium.com/Blahah/transrate.png?travis)][gemnasium]
-[![Code Climate](https://codeclimate.com/github/Blahah/transrate.png)][codeclimate]
-[![Coverage Status](https://coveralls.io/repos/Blahah/transrate/badge.png?branch=master)][coveralls]
+[![Gem Version](http://img.shields.io/gem/v/transrate.svg)][gem]
+![Downloads](http://img.shields.io/gem/dtv/transrate.svg)
+[![Build Status](http://img.shields.io/travis/Blahah/transrate/master.svg)][travis]
+[![Dependency Status](http://img.shields.io/gemnasium/Blahah/transrate.svg)][gemnasium]
+[![Code Climate](http://img.shields.io/codeclimate/github/Blahah/transrate.svg)][codeclimate]
+[![Coverage Status](http://img.shields.io/coveralls/Blahah/transrate.svg)][coveralls]
 
 [gem]: https://badge.fury.io/rb/transrate
 [travis]: https://travis-ci.org/Blahah/transrate
@@ -20,7 +21,7 @@ This software is being actively developed. Please be aware that there may be bug
 
 ## Citation
 
-Transrate is pre-publication academic software. If you use it, please cite the github repository and the DOI: [![DOI](https://zenodo.org/badge/3687/Blahah/transrate.png)](http://dx.doi.org/10.5281/zenodo.11037).
+Transrate is pre-publication academic software. If you use it, please cite the github repository and the DOI: [![DOI](https://zenodo.org/badge/3687/Blahah/transrate.png)](http://dx.doi.org/10.5281/zenodo.11039).
 
 ## Documentation
 

data/bin/transrate

@@ -46,12 +46,6 @@ opts = Trollop::options do
       :type => String
   opt :right, "right reads file in FASTQ format",
       :type => String
-  opt :insertsize, "mean insert size",
-      :default => 200,
-      :type => Integer
-  opt :insertsd, "insert size standard deviation",
-      :default => 50,
-      :type => Integer
   opt :threads, "number of threads to use",
       :default => 8,
       :type => Integer
@@ -98,6 +92,21 @@ if opts.reference && !File.exist?(opts.reference)
   raise IOError.new "Reference fasta file does not exist: #{opts.reference}"
 end
 
+if opts.left and opts.right
+  if opts.left.split(",").length != opts.right.split(",").length
+    msg = "Please provide the same number of left reads as right reads"
+    raise ArgumentError.new(msg)
+  end
+  opts.left.split(",").zip(opts.right.split(",")).each do |left,right|
+    if !File.exist?(left)
+      raise IOError.new "Left read fastq file does not exist: #{left}"
+    end
+    if !File.exist?(right)
+      raise IOError.new "Right read fastq file does not exist: #{right}"
+    end
+  end
+end
+
 if opts.profile
   logger.info "Starting profiler"
   RubyProf.start
@@ -201,13 +210,16 @@ opts.assembly.split(',').each do |assembly|
     logger.info "Comparative metrics done in #{Time.now - t0} seconds"
 
     logger.info "-" * report_width
+  else
+    logger.info "No reference provided, skipping comparative diagnostics"
+  end
+
+  if (opts.left && opts.right)
     score = transrater.assembly_score
     unless score.nil?
-      logger.info "OVERALL SCORE: #{score.to_f.round(2) * 100}%"
+      logger.info "TRANSRATE ASSEMBLY SCORE: #{score.round(2)}"
       logger.info "-" * report_width
     end
-  else
-    logger.info "No reference provided, skipping comparative diagnostics"
   end
 
   # write contig metrics to file for each contig

data/deps/deps.yaml

@@ -25,21 +25,56 @@ blastplus:
     64bit:
       macosx: ftp://ftp.ncbi.nlm.nih.gov/blast/executables/blast+/2.2.29/ncbi-blast-2.2.29+-universal-macosx.tar.gz
       linux: ftp://ftp.ncbi.nlm.nih.gov/blast/executables/blast+/2.2.29/ncbi-blast-2.2.29+-x64-linux.tar.gz
-bowtie2:
+snap:
   binaries:
-    - bowtie2
-    - bowtie2-align-l
-    - bowtie2-align-s
-    - bowtie2-build
-    - bowtie2-build-l
-    - bowtie2-build-s
-    - bowtie2-inspect
-    - bowtie2-inspect-l
-    - bowtie2-inspect-s
+    - snap
   version:
-    number: '2.2.3'
-    command: 'bowtie2 --version'
+    number: '1.0dev.50'
+    command: 'snap'
   url:
     64bit:
-      linux: http://downloads.sourceforge.net/project/bowtie-bio/bowtie2/2.2.3/bowtie2-2.2.3-linux-x86_64.zip
-      macosx: http://downloads.sourceforge.net/project/bowtie-bio/bowtie2/2.2.3/bowtie2-2.2.3-macos-x86_64.zip
+      linux: https://github.com/cboursnell/snap/raw/dev/bin/linux/snap.tar.gz
+      macosx: https://github.com/HibberdLab/snap/raw/dev/bin/macosx/snap.tar.gz
+samtools:
+  binaries:
+    - samtools
+    - bcftools
+  version:
+    number: '0.1.19'
+    command: 'samtools'
+  url:
+    64bit:
+      linux: https://github.com/cboursnell/samtools/raw/master/build/linux64.tar.gz
+      macosx: https://github.com/cboursnell/samtools/raw/master/build/osx64.tar.gz
+bam-read:
+  binaries:
+   - bam-read
+  version:
+    number: '0.3\.1'
+    command: 'bam-read'
+  url:
+    64bit:
+      linux: https://github.com/cboursnell/transrate-bam-read/raw/master/bin/linux/bam-read
+      macosx: https://github.com/Blahah/transrate-bam-read/raw/master/bin/macosx/bam-read
+  unpack: false
+bam-split:
+  binaries:
+   - bam-split
+  version:
+    number: '0.1'
+    command: 'bam-split'
+  url:
+    64bit:
+      linux: https://github.com/Blahah/transrate-bam-read/raw/master/bin/linux/bam-split
+      macosx: https://github.com/Blahah/transrate-bam-read/raw/master/bin/macosx/bam-split
+  unpack: false
+express:
+  binaries:
+    - express
+  version:
+    number: '1.5.1'
+    command: 'express --version'
+  url:
+    64bit:
+      linux: http://bio.math.berkeley.edu/eXpress/downloads/express-1.5.1/express-1.5.1-linux_x86_64.tgz
+      macosx: http://bio.math.berkeley.edu/eXpress/downloads/express-1.5.1/express-1.5.1-macosx_x86_64.tgz

data/ext/transrate/transrate.c

@@ -1,223 +1,257 @@
 #include "ruby.h"
 #include <stdlib.h>
+#include <math.h>
 
 // Defining a space for information and references about the module to be
 // stored internally
 VALUE Contig = Qnil;
+VALUE ReadMetrics = Qnil;
 VALUE Transrate = Qnil;
 
 // Prototype for the initialization method - Ruby calls this, not you
 void Init_transrate();
 
 // methods are prefixed by 'method_' here
-//VALUE TestInit(VALUE, VALUE, VALUE, VALUE, VALUE);
+// contig
 VALUE method_composition(VALUE, VALUE);
 VALUE method_base_count(VALUE,VALUE);
 VALUE method_dibase_count(VALUE,VALUE);
 VALUE method_kmer_count(VALUE,VALUE,VALUE);
 VALUE method_longest_orf(VALUE, VALUE);
+// read_metrics
 
 int * base_counts;
 int * dibase_counts;
 
 // The initialization method for this module
 void Init_transrate() {
-    Transrate = rb_define_module("Transrate");
-    // VALUE rb_define_class_under(VALUE outer, const char *name, VALUE super)
-    Contig = rb_define_class_under(Transrate, "Contig", rb_cObject);
-    // rb_define_method(Contig, "initialize", TestInit, 2);
-    rb_define_method(Contig, "composition", method_composition, 1);
-    rb_define_method(Contig, "base_count", method_base_count, 1);
-    rb_define_method(Contig, "dibase_count", method_dibase_count, 1);
-    rb_define_method(Contig, "kmer_count", method_kmer_count, 2);
-    rb_define_method(Contig, "longest_orf", method_longest_orf, 1);
+  Transrate = rb_define_module("Transrate");
+  Contig = rb_define_class_under(Transrate, "Contig", rb_cObject);
+  ReadMetrics = rb_define_class_under(Transrate, "ReadMetrics", rb_cObject);
+  // contig
+  rb_define_method(Contig, "composition", method_composition, 1);
+  rb_define_method(Contig, "base_count", method_base_count, 1);
+  rb_define_method(Contig, "dibase_count", method_dibase_count, 1);
+  rb_define_method(Contig, "kmer_count", method_kmer_count, 2);
+  rb_define_method(Contig, "longest_orf", method_longest_orf, 1);
+  // ReadMetrics
 }
 
 VALUE method_composition(VALUE self, VALUE _seq) {
-    int i,len, idx;
-    char * seq;
-    char base;
-    char prevbase;
-    seq = StringValueCStr(_seq);
-    len = RSTRING_LEN(_seq);
-    base_counts = malloc(5 * sizeof(int));
-    dibase_counts = malloc(25 * sizeof(int));
+  int i, len, idx;
+  char * seq;
+  char base;
+  char prevbase;
+  seq = StringValueCStr(_seq);
+  len = RSTRING_LEN(_seq);
+  base_counts = malloc(5 * sizeof(int));
+  dibase_counts = malloc(25 * sizeof(int));
 
-    for (i=0; i < 5; i++) {
-        base_counts[i]=0;
+  for (i=0; i < 5; i++) {
+    base_counts[i]=0;
+  }
+  for (i=0; i < 25; i++) {
+    dibase_counts[i]=0;
+  }
+  for (i=0; i < len; i++) {
+    base = seq[i];
+    switch (base) {
+      case 'A': {
+        idx=0;
+        break;
+      }
+      case 'C': {
+        idx=1;
+        break;
+      }
+      case 'G': {
+        idx=2;
+        break;
+      }
+      case 'T': {
+        idx=3;
+        break;
+      }
+      default: {
+        idx=4;
+        break;
+      }
     }
-    for (i=0; i < 25; i++) {
-        dibase_counts[i]=0;
-    }
-    for (i=0; i < len; i++) {
-        base = seq[i];
-        switch (base) {
-            case 'A': {
-                idx=0;
-                break;
-            }
-            case 'C': {
-                idx=1;
-                break;
-            }
-            case 'G': {
-                idx=2;
-                break;
-            }
-            case 'T': {
-                idx=3;
-                break;
-            }
-            default: {
-                idx=4;
-                break;
-            }
-        }
-        base_counts[idx]++;
+    base_counts[idx]++;
 
-        if (i > 0) {
-            prevbase = seq[i-1];
-            switch (prevbase) {
-                case 'A': {
-                    idx=idx;
-                    break;
-                }
-                case 'C': {
-                    idx=idx+5;
-                    break;
-                }
-                case 'G': {
-                    idx=idx+10;
-                    break;
-                }
-                case 'T': {
-                    idx=idx+15;
-                    break;
-                }
-                default: {
-                    idx=idx+20;
-                    break;
-                }
-            }
-            dibase_counts[idx]++;
+    if (i > 0) {
+      prevbase = seq[i-1];
+      switch (prevbase) {
+        case 'A': {
+          idx=idx;
+          break;
+        }
+        case 'C': {
+          idx=idx+5;
+          break;
+        }
+        case 'G': {
+          idx=idx+10;
+          break;
+        }
+        case 'T': {
+          idx=idx+15;
+          break;
         }
+        default: {
+          idx=idx+20;
+          break;
+        }
+      }
+      dibase_counts[idx]++;
     }
-    return INT2NUM(0);
+  }
+  return INT2NUM(0);
 }
 
 VALUE method_dibase_count(VALUE self, VALUE idx) {
-    return INT2NUM(dibase_counts[NUM2INT(idx)]);
+  return INT2NUM(dibase_counts[NUM2INT(idx)]);
 }
 
 VALUE method_base_count(VALUE self, VALUE idx) {
-    return INT2NUM(base_counts[NUM2INT(idx)]);
+  return INT2NUM(base_counts[NUM2INT(idx)]);
 }
 
 VALUE method_kmer_count(VALUE self, VALUE _k, VALUE _s) {
-    int n, i, start, k, len, h, size = 0;
-    char * c_str;
-    char base;
-    len = RSTRING_LEN(_s);
-    c_str = StringValueCStr(_s);
-    k = NUM2INT(_k);
-    size = 1;
-    for(h=0;h<k;h++) {
-        size *= 4;
-    }
-    short set[size];
-    for(start=0;start<size;start++) {
-        set[start]=0;
-    }
-    for(start=0; start<len-k+1; start++) {
-        i = 0;
-        h = 0;
-        n = 0;
-        for(i = start; i < start+k; i++) {
-            base = c_str[i];
-            switch (base) {
-                case 'A': {
-                    h = h << 2;
-                    h += 0;
-                    break;
-                }
-                case 'C': {
-                    h = h << 2;
-                    h += 1;
-                    break;
-                }
-                case 'G': {
-                    h = h << 2;
-                    h += 2;
-                    break;
-                }
-                case 'T': {
-                    h = h << 2;
-                    h += 3;
-                    break;
-                }
-                default: {
-                    n++;
-                    break;
-                }
-            }
+  int n, i, start, k, len, h, size = 0;
+  char * c_str;
+  char base;
+  len = RSTRING_LEN(_s);
+  c_str = StringValueCStr(_s);
+  k = NUM2INT(_k);
+  size = 1;
+  for(h=0;h<k;h++) {
+      size *= 4;
+  }
+  short set[size];
+  for(start=0;start<size;start++) {
+      set[start]=0;
+  }
+  for(start=0; start<len-k+1; start++) {
+    i = 0;
+    h = 0;
+    n = 0;
+    for(i = start; i < start+k; i++) {
+      base = c_str[i];
+      switch (base) {
+        case 'A': {
+          h = h << 2;
+          h += 0;
+          break;
         }
-        if (n==0) {
-            set[h] += 1;
+        case 'C': {
+          h = h << 2;
+          h += 1;
+          break;
         }
-    }
-    i = 0; // count how many in array are set //
-    for(start = 0; start < size; start++) {
-        if (set[start]>0) {
-            i++;
+        case 'G': {
+          h = h << 2;
+          h += 2;
+          break;
+        }
+        case 'T': {
+          h = h << 2;
+          h += 3;
+          break;
         }
+        default: {
+          n++;
+          break;
+        }
+      }
+    }
+    if (n==0) {
+      set[h] += 1;
+    }
+  }
+  i = 0; // count how many in array are set //
+  for(start = 0; start < size; start++) {
+    if (set[start]>0) {
+      i++;
     }
-    return INT2NUM(i);
+  }
+  return INT2NUM(i);
 }
 
 // takes in a string and calculates the longest open reading frame
 // in any of the 6 frames
 // an open reading frame is defined as the number of bases between
-// either the start of the sequence or a stop codon and either the
+// either the start of the sequence or a start codon and either the
 // end of the sequence or a stop codon
-VALUE method_longest_orf(VALUE self, VALUE _s) {
-    int i,sl,longest=0;
-    int len[6];
-    char * c_str;
 
-    sl = RSTRING_LEN(_s);
-    c_str = StringValueCStr(_s);
-    for (i=0;i<6;i++) {
-        len[i]=0;
-    }
-    for (i=0;i<sl-2;i++) {
-        if (c_str[i]=='T' &&
-        ((c_str[i+1]=='A' && c_str[i+2]=='G') ||
-        (c_str[i+1]=='A' && c_str[i+2]=='A') ||
-        (c_str[i+1]=='G' && c_str[i+2]=='A'))) {
-            if (len[i%3] > longest) {
-                longest = len[i%3];
-            }
-            len[i%3]=0;
-        } else {
-            len[i%3]++;
+VALUE method_longest_orf(VALUE self, VALUE _str) {
+  int i,sl,longest=0;
+  int len[3];
+  char * str;
+  sl = RSTRING_LEN(_str);
+  str = StringValueCStr(_str);
+  for (i=0;i<3;i++) {
+    len[i]=0;
+  }
+  for(i=0;i<sl-2;i++) {
+    if (str[i]=='A' && str[i+1]=='T' && str[i+2]=='G') { //Methionine
+      if (len[i%3]>=0) {
+        len[i%3]++;
+      } else {
+        len[i%3]=1;
+      }
+    } else {
+      if (str[i]=='T' &&
+        ((str[i+1]=='A' && str[i+2]=='G') ||   //amber
+        (str[i+1]=='A' && str[i+2]=='A') ||    //ochre   stops
+        (str[i+1]=='G' && str[i+2]=='A'))) {   //umber
+        if (len[i%3]>longest) {
+          longest = len[i%3];
         }
-        if (c_str[i+2]=='A' &&
-        ((c_str[i]=='C' && c_str[i+1]=='T') ||
-        (c_str[i]=='T' && c_str[i+1]=='T') ||
-        (c_str[i]=='T' && c_str[i+1]=='C'))) {
-            if (len[3+i%3] > longest) {
-                longest = len[3+i%3];
-            }
-            len[3+i%3]=0;
-        } else {
-            len[3+i%3]++;
+        len[i%3]=-1;
+      } else { // any other codon
+        if (len[i%3]>=0) {
+          len[i%3]++;
         }
+      }
     }
+  }
+  for(i=0;i<3;i++) {
     if (len[i%3] > longest) {
-        longest = len[i%3];
+      longest = len[i%3];
+    }
+  }
+  for (i=0;i<3;i++) {
+    len[i]=0;
+  }
+  for(i=sl-1;i>=2;i--) {
+    if (str[i]=='T' && str[i-1]=='A' && str[i-2]=='C') { //Methionine
+      if (len[i%3]>=0) {
+        len[i%3]++;
+      } else {
+        len[i%3]=1;
+      }
+    } else {
+      if (str[i]=='A' &&
+        ((str[i-1]=='T' && str[i-2]=='C') ||   //amber
+        (str[i-1]=='T' && str[i-2]=='T') ||    //ochre   stops
+        (str[i-1]=='C' && str[i-2]=='T'))) {   //umber
+        if (len[i%3]>longest) {
+          longest = len[i%3];
+        }
+        len[i%3]=-1;
+      } else { // any other codon
+        if (len[i%3]>=0) {
+          len[i%3]++;
+        }
+      }
     }
-    if (len[3+i%3] > longest) {
-        longest = len[3+i%3];
+  }
+  for(i=0;i<3;i++) {
+    if (len[i%3] > longest) {
+      longest = len[i%3];
     }
-    return INT2NUM(longest);
-}
+  }
+  return INT2NUM(longest);
+}
+