minimap2 0.0.4 → 0.2.23.1

data/ext/minimap2/tex/ngmlr.eval

@@ -0,0 +1,38 @@
+Q	60	23616	0	0.000000000
+Q	45	3520	1	0.000036851
+Q	41	1840	1	0.000069023
+Q	37	328	2	0.000136500
+Q	36	276	1	0.000169033
+Q	35	480	1	0.000199601
+Q	33	375	2	0.000262855
+Q	31	178	2	0.000326659
+Q	30	153	5	0.000487551
+Q	29	200	1	0.000516696
+Q	27	100	3	0.000611601
+Q	26	93	3	0.000706056
+Q	25	75	2	0.000768393
+Q	24	82	1	0.000798314
+Q	23	80	6	0.000987387
+Q	22	71	6	0.001175835
+Q	21	76	7	0.001394921
+Q	20	63	9	0.001676897
+Q	19	55	4	0.001800322
+Q	18	62	8	0.002048987
+Q	17	55	7	0.002265718
+Q	16	60	10	0.002575539
+Q	15	82	9	0.002850877
+Q	14	67	7	0.003063745
+Q	13	62	11	0.003401042
+Q	12	64	13	0.003799084
+Q	11	56	5	0.003947900
+Q	10	58	17	0.004468303
+Q	9	70	22	0.005139796
+Q	8	23	9	0.005414604
+Q	7	41	17	0.005933068
+Q	6	42	18	0.006480881
+Q	5	33	9	0.006751757
+Q	4	29	9	0.007022948
+Q	3	27	15	0.007478764
+Q	2	23	10	0.007781024
+Q	1	9	2	0.007840364
+Q	0	13	8	0.008083105

data/ext/minimap2/tex/roc.gp

@@ -0,0 +1,60 @@
+set t po eps enh co so "Helvetica,26"
+
+set style line 1 lt 1 pt 1 lc rgb "#e41a1c" lw 2;
+set style line 2 lt 1 pt 2 lc rgb "#377eb8" lw 2;
+set style line 3 lt 1 pt 3 lc rgb "#4daf4a" lw 2;
+set style line 4 lt 1 pt 4 lc rgb "#984ea3" lw 2;
+set style line 5 lt 1 pt 6 lc rgb "#ff7f00" lw 2;
+set style line 6 lt 1 pt 8 lc rgb "#f781bf" lw 2;
+
+set out "roc-color.eps"
+
+set pointsize 2.0
+set size 1.59,1.04
+set multiplot layout 1,2
+
+set label "(a)" at graph -0.245,1.06 font "Helvetica-bold,40"
+set xlab "Error rate of mapped PacBio reads"
+set ylab "Fraction of mapped reads" off +1.8
+set ytics 0.02
+set yran [0.9:1]
+
+set size 0.8,1
+set log x
+set format x "10^{%L}"
+set key bot right
+plot "<./eval2roc.pl blasr-mc.eval" u 2:3 t "blasr-mc" w lp ls 4, \
+     "<./eval2roc.pl bwa.eval" u 2:3 t "bwa-mem" w lp ls 2, \
+     "<./eval2roc.pl graphmap.eval" u 2:3 t "graphmap" w lp ls 3, \
+     "<./eval2roc.pl minialign.eval" u 2:3 t "minialign" w lp ls 1, \
+     "<./eval2roc.pl mm2.eval" u 2:3 t "minimap2" w lp ls 6, \
+     "<./eval2roc.pl ngmlr.eval" u 2:3 t "ngm-lr" w lp ls 5
+unset label
+
+set origin 0.8,0
+set size 0.79,1
+set label "(b)" at graph -0.245,1.06 font "Helvetica-bold,40"
+set xlab "Error rate of mapped short reads"
+
+set key top left
+plot "<./eval2roc.pl -n2e7 bowtie2-s3.sam.eval" u 2:3 t "bowtie2" w lp ls 5, \
+	 "<./eval2roc.pl -n2e7 bwa-s3.sam.eval" u 2:3 t "bwa-mem" w lp ls 2, \
+	 "<./eval2roc.pl -n2e7 mm2-s3.sam.eval" u 2:3 t "minimap2" w lp ls 6, \
+	 "<./eval2roc.pl -n2e7 snap-s3.sam.eval" u 2:3 t "snap" w lp ls 3
+
+#unset log
+#unset format
+#unset key
+#set log y
+#set ylab "Accumulative mapping error rate" off +0
+#set xlab "Mapping quality"
+#set yran [1e-5:0.1]
+#set ytics 1e-5,0.1
+#set format y "10^{%L}"
+#set xran [60:0] reverse
+#plot "<./eval2roc.pl blasr-mc.eval" u 1:2 w lp ls 4, \
+#     "<./eval2roc.pl bwa.eval" u 1:2 t "bwa-mem" w lp ls 2, \
+#     "<./eval2roc.pl graphmap.eval" u 1:2 t "graphmap" w lp ls 3, \
+#     "<./eval2roc.pl minialign.eval" u 1:2 t "minialign" w lp ls 1, \
+#     "<./eval2roc.pl mm2.eval" u 1:2 t "minimap2" w lp ls 6, \
+#     "<./eval2roc.pl ngmlr.eval" u 1:2 t "ngm-lr" w lp ls 5

data/ext/minimap2/tex/snap-s3.sam.eval

@@ -0,0 +1,62 @@
+Q	60	18993268	10320	0.000543350	18993268
+Q	59	33156	216	0.000553756	19026424
+Q	58	29982	295	0.000568365	19056406
+Q	57	9412	278	0.000582666	19065818
+Q	56	11012	228	0.000594281	19076830
+Q	55	9968	235	0.000606283	19086798
+Q	54	8602	292	0.000621301	19095400
+Q	53	6094	259	0.000634662	19101494
+Q	52	5026	257	0.000647946	19106520
+Q	51	4278	224	0.000659522	19110798
+Q	50	3682	178	0.000668708	19114480
+Q	49	2750	156	0.000676772	19117230
+Q	48	2314	112	0.000682548	19119544
+Q	47	2056	96	0.000687495	19121600
+Q	46	1658	62	0.000690677	19123258
+Q	45	1492	74	0.000694493	19124750
+Q	44	1150	56	0.000697379	19125900
+Q	43	1062	48	0.000699850	19126962
+Q	42	976	60	0.000702951	19127938
+Q	41	884	36	0.000704800	19128822
+Q	40	708	52	0.000707493	19129530
+Q	39	870	26	0.000708819	19130400
+Q	38	598	26	0.000710156	19130998
+Q	37	542	34	0.000711913	19131540
+Q	36	846	50	0.000714495	19132386
+Q	35	590	50	0.000717087	19132976
+Q	34	550	42	0.000719261	19133526
+Q	33	2174	66	0.000722628	19135700
+Q	32	876	86	0.000727089	19136576
+Q	31	638	104	0.000732500	19137214
+Q	30	1718	196	0.000742675	19138932
+Q	29	91022	968	0.000789497	19229954
+Q	28	12864	781	0.000829556	19242818
+Q	27	5806	427	0.000851489	19248624
+Q	26	25274	728	0.000888144	19273898
+Q	25	7418	680	0.000923070	19281316
+Q	24	11800	701	0.000958839	19293116
+Q	23	57328	3933	0.001159250	19350444
+Q	22	7662	846	0.001202494	19358106
+Q	21	5924	617	0.001233989	19364030
+Q	20	4623	574	0.001263330	19368653
+Q	19	4988	942	0.001311627	19373641
+Q	18	3968	793	0.001352282	19377609
+Q	17	3630	681	0.001387166	19381239
+Q	16	2921	513	0.001413422	19384160
+Q	15	2716	424	0.001435095	19386876
+Q	14	2366	365	0.001453744	19389242
+Q	13	2169	412	0.001474828	19391411
+Q	12	2077	360	0.001493233	19393488
+Q	11	2016	441	0.001515815	19395504
+Q	10	2292	738	0.001553682	19397796
+Q	9	4165	1832	0.001647772	19401961
+Q	8	3963	1862	0.001743385	19405924
+Q	7	3927	1793	0.001835408	19409851
+Q	6	3572	1639	0.001919497	19413423
+Q	5	3270	1533	0.001998126	19416693
+Q	4	3046	1610	0.002080718	19419739
+Q	3	251447	125550	0.008436553	19671186
+Q	2	24390	13537	0.009113417	19695576
+Q	1	124406	86780	0.013434624	19819982
+Q	0	171254	153874	0.021016609	19991236
+U	8764

data/ext/minimap2.patch

@@ -0,0 +1,19 @@
+--- Makefile.org	2021-05-27 15:45:11.993128205 +0900
++++ Makefile	2021-05-27 15:46:02.320569154 +0900
+@@ -1,9 +1,9 @@
+-CFLAGS=		-g -Wall -O2 -Wc++-compat #-Wextra
++CFLAGS=		-g -Wall -O2 -Wc++-compat -fPIC #-Wextra
+ CPPFLAGS=	-DHAVE_KALLOC
+ INCLUDES=
+ OBJS=		kthread.o kalloc.o misc.o bseq.o sketch.o sdust.o options.o index.o \
+ 			lchain.o align.o hit.o seed.o map.o format.o pe.o esterr.o splitidx.o \
+-			ksw2_ll_sse.o
++			ksw2_ll_sse.o cmappy.o
+ PROG=		minimap2
+ PROG_EXTRA=	sdust minimap2-lite
+ LIBS=		-lm -lz -lpthread
+@@ -130,3 +130,4 @@ sdust.o: kalloc.h kdq.h kvec.h sdust.h
+ seed.o: mmpriv.h minimap.h bseq.h kseq.h kalloc.h ksort.h
+ sketch.o: kvec.h kalloc.h mmpriv.h minimap.h bseq.h kseq.h
+ splitidx.o: mmpriv.h minimap.h bseq.h kseq.h
++cmappy.o: cmappy.h

data/lib/minimap2/aligner.rb

@@ -4,11 +4,21 @@ module Minimap2
   class Aligner
     attr_reader :idx_opt, :map_opt, :index
 
-    # Create a new aligner
+    # Create a new aligner.
     #
     # @param fn_idx_in [String] index or sequence file name.
     # @param seq [String] a single sequence to index.
     # @param preset [String] minimap2 preset.
+    #   * map-pb : PacBio CLR genomic reads
+    #   * map-ont : Oxford Nanopore genomic reads
+    #   * map-hifi : PacBio HiFi/CCS genomic reads (v2.19 or later)
+    #   * asm20 : PacBio HiFi/CCS genomic reads (v2.18 or earlier)
+    #   * sr : short genomic paired-end reads
+    #   * splice : spliced long reads (strand unknown)
+    #   * splice:hq : Final PacBio Iso-seq or traditional cDNA
+    #   * asm5 : intra-species asm-to-asm alignment
+    #   * ava-pb : PacBio read overlap
+    #   * ava-ont : Nanopore read overlap
     # @param k [Integer] k-mer length, no larger than 28.
     # @param w [Integer] minimizer window size, no larger than 255.
     # @param min_cnt [Integer] mininum number of minimizers on a chain.
@@ -101,6 +111,7 @@ module Minimap2
     end
 
     # Explicitly releases the memory of the index object.
+
     def free_index
       FFI.mm_idx_destroy(index) unless index.null?
     end
@@ -184,10 +195,10 @@ module Minimap2
       alignments
     end
 
-    # retrieve a subsequence from the index.
-    # @params name
-    # @params start
-    # @params stop
+    # Retrieve a subsequence from the index.
+    # @param name
+    # @param start
+    # @param stop
 
     def seq(name, start = 0, stop = 0x7fffffff)
       lp = ::FFI::MemoryPointer.new(:int)

data/lib/minimap2/alignment.rb

@@ -1,7 +1,7 @@
 # frozen_string_literal: true
 
 module Minimap2
-  # Alignment result
+  # Alignment result.
   #
   # @!attribute ctg
   #   @return [String] name of the reference sequence the query is mapped to.
@@ -73,17 +73,21 @@ module Minimap2
       @cs           = cs
       @md           = md
 
-      @cigar_str = cigar.map { |x| x[0].to_s + 'MIDNSH'[x[1]] }.join
+      @cigar_str = cigar.map { |x| x[0].to_s + FFI::CIGAR_STR[x[1]] }.join
     end
 
     def primary?
       @primary == 1
     end
 
+    # Convert Alignment to hash.
+
     def to_h
       self.class.keys.map { |k| [k, __send__(k)] }.to_h
     end
 
+    # Convert to the PAF format without the QueryName and QueryLength columns.
+
     def to_s
       strand = if @strand.positive?
                  '+'

data/lib/minimap2/ffi/constants.rb

@@ -3,45 +3,60 @@
 module Minimap2
   module FFI
     # flags
-    NO_DIAG        = 0x001 # no exact diagonal hit
-    NO_DUAL        = 0x002 # skip pairs where query name is lexicographically larger than target name
-    CIGAR          = 0x004
-    OUT_SAM        = 0x008
-    NO_QUAL        = 0x010
-    OUT_CG         = 0x020
-    OUT_CS         = 0x040
-    SPLICE         = 0x080 # splice mode
-    SPLICE_FOR     = 0x100 # match GT-AG
-    SPLICE_REV     = 0x200 # match CT-AC, the reverse complement of GT-AG
-    NO_LJOIN       = 0x400
-    OUT_CS_LONG    = 0x800
-    SR             = 0x1000
-    FRAG_MODE      = 0x2000
-    NO_PRINT_2ND   = 0x4000
-    TWO_IO_THREADS = 0x8000 # Translator's Note. MM_F_2_IO_THREADS. Constants starting with numbers cannot be defined.
-    LONG_CIGAR     = 0x10000
-    INDEPEND_SEG   = 0x20000
-    SPLICE_FLANK   = 0x40000
-    SOFTCLIP       = 0x80000
-    FOR_ONLY       = 0x100000
-    REV_ONLY       = 0x200000
-    HEAP_SORT      = 0x400000
-    ALL_CHAINS     = 0x800000
-    OUT_MD         = 0x1000000
-    COPY_COMMENT   = 0x2000000
-    EQX            = 0x4000000 # use =/X instead of M
-    PAF_NO_HIT     = 0x8000000 # output unmapped reads to PAF
-    NO_END_FLT     = 0x10000000
-    HARD_MLEVEL    = 0x20000000
-    SAM_HIT_ONLY   = 0x40000000
-
-    HPC            = 0x1
-    NO_SEQ         = 0x2
-    NO_NAME        = 0x4
-
-    IDX_MAGIC      = "MMI\2"
-
-    MAX_SEG        = 255
+    NO_DIAG          = 0x001 # no exact diagonal hit
+    NO_DUAL          = 0x002 # skip pairs where query name is lexicographically larger than target name
+    CIGAR            = 0x004
+    OUT_SAM          = 0x008
+    NO_QUAL          = 0x010
+    OUT_CG           = 0x020
+    OUT_CS           = 0x040
+    SPLICE           = 0x080 # splice mode
+    SPLICE_FOR       = 0x100 # match GT-AG
+    SPLICE_REV       = 0x200 # match CT-AC, the reverse complement of GT-AG
+    NO_LJOIN         = 0x400
+    OUT_CS_LONG      = 0x800
+    SR               = 0x1000
+    FRAG_MODE        = 0x2000
+    NO_PRINT_2ND     = 0x4000
+    TWO_IO_THREADS   = 0x8000 # Translator's Note. MM_F_2_IO_THREADS. Constants starting with numbers cannot be defined.
+    LONG_CIGAR       = 0x10000
+    INDEPEND_SEG     = 0x20000
+    SPLICE_FLANK     = 0x40000
+    SOFTCLIP         = 0x80000
+    FOR_ONLY         = 0x100000
+    REV_ONLY         = 0x200000
+    HEAP_SORT        = 0x400000
+    ALL_CHAINS       = 0x800000
+    OUT_MD           = 0x1000000
+    COPY_COMMENT     = 0x2000000
+    EQX              = 0x4000000 # use =/X instead of M
+    PAF_NO_HIT       = 0x8000000 # output unmapped reads to PAF
+    NO_END_FLT       = 0x10000000
+    HARD_MLEVEL      = 0x20000000
+    SAM_HIT_ONLY     = 0x40000000
+    RMQ              = 0x80000000  # LL
+    QSTRAND          = 0x100000000 # LL
+    NO_INV           = 0x200000000
+
+    HPC              = 0x1
+    NO_SEQ           = 0x2
+    NO_NAME          = 0x4
+
+    IDX_MAGIC        = "MMI\2"
+
+    MAX_SEG          = 255
+
+    CIGAR_MATCH      = 0
+    CIGAR_INS        = 1
+    CIGAR_DEL        = 2
+    CIGAR_N_SKIP     = 3
+    CIGAR_SOFTCLIP   = 4
+    CIGAR_HARDCLIP   = 5
+    CIGAR_PADDING    = 6
+    CIGAR_EQ_MATCH   = 7
+    CIGAR_X_MISMATCH = 8
+
+    CIGAR_STR        = 'MIDNSHP=XB'
 
     # emulate 128-bit integers
     class MM128 < ::FFI::Struct
@@ -86,6 +101,7 @@ module Minimap2
         :min_cnt,              :int,     # min number of minimizers on each chain
         :min_chain_score,      :int,     # min chaining score
         :chain_gap_scale,      :float,
+        :chain_skip_scale,     :float,
         :rmq_size_cap,         :int,
         :rmq_inner_dist,       :int,
         :rmq_rescue_size,      :int,
@@ -112,14 +128,18 @@ module Minimap2
         :anchor_ext_len,       :int,
         :anchor_ext_shift,     :int,
         :max_clip_ratio,       :float,   # drop an alignment if BOTH ends are clipped above this ratio
+        :rank_min_len,         :int,
+        :rank_frac,            :float,
         :pe_ori,               :int,
         :pe_bonus,             :int,
         :mid_occ_frac,         :float,   # only used by mm_mapopt_update(); see below
+        :q_occ_frac,           :float,
         :min_mid_occ,          :int32_t,
         :mid_occ,              :int32_t, # ignore seeds with occurrences above this threshold
         :max_occ,              :int32_t,
         :mini_batch_size,      :int64_t, # size of a batch of query bases to process in parallel
         :max_sw_mat,           :int64_t,
+        :cap_kalloc,           :int64_t,
         :split_prefix,         :string
     end
 
@@ -170,7 +190,7 @@ module Minimap2
         :n_ambi_trans_strand, :uint32,
         :n_cigar,             :uint32
 
-      bitfields :n_ambi_trans_strand,
+      bit_field :n_ambi_trans_strand,
                 :n_ambi, 30,      # number of ambiguous bases
                 :trans_strand, 2  # transcript strand: 0 for unknown, 1 for +, 2 for -
 
@@ -202,19 +222,20 @@ module Minimap2
         :div,    :float,
         :p,      Extra.ptr
 
-      bitfields :fields,
-                :mapq,        8,
-                :split,       2,
-                :rev,         1,
-                :inv,         1,
-                :sam_pri,     1,
-                :proper_frag, 1,
-                :pe_thru,     1,
-                :seg_split,   1,
-                :seg_id,      8,
-                :split_inv,   1,
-                :is_alt,      1,
-                :dummy,       6
+      bit_field :fields,
+                :mapq,            8,
+                :split,           2,
+                :rev,             1,
+                :inv,             1,
+                :sam_pri,         1,
+                :proper_frag,     1,
+                :pe_thru,         1,
+                :seg_split,       1,
+                :seg_id,          8,
+                :split_inv,       1,
+                :is_alt,          1,
+                :strand_retained, 1,
+                :dummy,           5
     end
 
     # memory buffer for thread-local storage during mapping

data/lib/minimap2/ffi/functions.rb

@@ -2,6 +2,11 @@
 
 module Minimap2
   module FFI
+    attach_function \
+      :main,
+      %i[int pointer],
+      :int
+
     attach_function \
       :mm_set_opt_raw, :mm_set_opt,
       [:pointer, IdxOpt.by_ref, MapOpt.by_ref],

data/lib/minimap2/ffi.rb

@@ -1,8 +1,7 @@
 # frozen_string_literal: true
 
 # bit fields
-require_relative 'ffi_helper'
-
+require 'ffi/bit_struct'
 module Minimap2
   # Native APIs
   module FFI

data/lib/minimap2/version.rb

@@ -1,5 +1,6 @@
 # frozen_string_literal: true
 
 module Minimap2
-  VERSION = '0.0.4'
+  # Minimap2-2.23 (r1111)
+  VERSION = '0.2.23.1'
 end

data/lib/minimap2.rb

@@ -3,9 +3,6 @@
 # dependencies
 require 'ffi'
 
-# bit fields
-require_relative 'minimap2/ffi_helper'
-
 # modules
 require_relative 'minimap2/aligner'
 require_relative 'minimap2/alignment'
@@ -34,30 +31,66 @@ module Minimap2
 
   # methods from mappy
   class << self
-    # read fasta/fastq file
+
+    # Execute minimap2 comannd with given options.
+    # @overload  execute(arg0,arg1,...)
+    # @param [String] arg minimap2 command option.
+    # @example Get minimap2 version
+    #   Minimap2.execute('--version')
+
+    def Minimap2.execute(*rb_argv)
+      str_ptrs = []
+      # First argument is the program name.
+      str_ptrs << ::FFI::MemoryPointer.from_string('minimap2')
+      rb_argv.each do |arg|
+        arg.to_s.split(/\s+/).each do |s|
+          str_ptrs << ::FFI::MemoryPointer.from_string(s)
+        end
+      end
+      strptrs << nil
+
+      # Load all the pointers into a native memory block
+      argv = ::FFI::MemoryPointer.new(:pointer, strptrs.length)
+      strptrs.each_with_index do |p, i|
+        argv[i].put_pointer(0,  p)
+      end
+
+      FFI.main(strptrs.length - 1, argv)
+    end
+
+    # Set verbosity level.
+    # @param [Integer] level
+
+    def verbose(level = -1)
+      FFI.mm_verbose_level(level)
+    end
+
+    # Read fasta/fastq file.
     # @param [String] file_path
-    # @param [Boolean] read_comment If false or nil, the comment will not be read.
+    # @param [Boolean] comment If True, the comment will be read.
     # @yield [name, seq, qual, comment]
-    # Note: You can also use a generic library such as BioRuby instead of this method.
+    # @return [Enumerator] enum Retrun Enumerator if not block given.
+    # Note: You can BioRuby instead of this method.
 
-    def fastx_read(file_path, read_comment = false)
+    def fastx_read(file_path, comment: false, &block)
       path = File.expand_path(file_path)
+
+      # raise error in Ruby because ks.null? is false even if file not exist.
+      raise ArgumentError, "File not found: #{path}" unless File.exist?(path)
+
       ks = FFI.mm_fastx_open(path)
-      while FFI.kseq_read(ks) >= 0
-        qual = ks[:qual][:s] if (ks[:qual][:l]).positive?
-        name = ks[:name][:s]
-        seq  = ks[:seq][:s]
-        if read_comment
-          comment = ks[:comment][:s] if (ks[:comment][:l]).positive?
-          yield [name, seq, qual, comment]
-        else
-          yield [name, seq, qual]
+
+      if block_given?
+        fastx_each(ks, comment, &block)
+      else
+        Enumerator.new do |y|
+          # rewind not work
+          fastx_each(ks, comment) { |r| y << r }
         end
       end
-      FFI.mm_fastx_close(ks)
     end
 
-    # reverse complement sequence
+    # Reverse complement sequence.
     # @param [String] seq
     # @return [string] seq
 
@@ -68,11 +101,23 @@ module Minimap2
       FFI.mappy_revcomp(l, bseq)
     end
 
-    # set verbosity level
-    # @param [Integer] level
+    private
 
-    def verbose(level = -1)
-      FFI.mm_verbose_level(level)
+    def fastx_each(ks, comment)
+      yield fastx_next(ks, comment) while FFI.kseq_read(ks) >= 0
+      FFI.mm_fastx_close(ks)
+    end
+
+    def fastx_next(ks, read_comment)
+      qual = ks[:qual][:s] if (ks[:qual][:l]).positive?
+      name = ks[:name][:s]
+      seq  = ks[:seq][:s]
+      if read_comment
+        comment = ks[:comment][:s] if (ks[:comment][:l]).positive?
+        [name, seq, qual, comment]
+      else
+        [name, seq, qual]
+      end
     end
   end
 end