Merge branch 'master' into featurecounts_strand

.github/workflows/main.yml

@@ -37,6 +37,8 @@ jobs:
 
       - name: Setup Snakemake environment
         run: |
+           # ensure that mamba is happy to write into the cache
+           sudo chown -R runner:docker /usr/share/miniconda/pkgs/cache
            conda install -c conda-forge mamba --quiet
            export PATH="/usr/share/miniconda/bin:$PATH"
            mamba create -c bioconda -c conda-forge --quiet -y --name snakemake snakemake-minimal pytest

.github/workflows/qc.yml

@@ -33,9 +33,11 @@ jobs:
 
       - name: Setup Snakemake environment
         run: |
-           conda install -c conda-forge mamba --quiet
-           export PATH="/usr/share/miniconda/bin:$PATH"
-           mamba create -c bioconda -c conda-forge --quiet -y --name snakemake snakemake-minimal
+          # ensure that mamba is happy to write into the cache
+          sudo chown -R runner:docker /usr/share/miniconda/pkgs/cache
+          conda install -c conda-forge mamba --quiet
+          export PATH="/usr/share/miniconda/bin:$PATH"
+          mamba create -c bioconda -c conda-forge --quiet -y --name snakemake snakemake-minimal
       
       - name: Fetch master
         run: |

bio/dragmap/align/environment.yaml

@@ -0,0 +1,9 @@
+channels:
+  - bioconda
+  - conda-forge
+  - defaults
+dependencies:
+  - dragmap =1.2
+  - samtools =1.14
+  - picard =2.26
+  - snakemake-wrapper-utils =0.3

bio/dragmap/align/meta.yaml

@@ -0,0 +1,13 @@
+name: Dragmap
+description: |
+  Map reads with Dragmap.
+url: https://github.com/Illumina/DRAGMAP
+authors:
+  - Filipe G. Vieira
+input:
+  - FASTQ file(s)
+  - reference hash table
+output:
+  - SAM/BAM/CRAM file
+notes: |
+  * The `extra` param allows for additional program arguments.

bio/dragmap/align/test/Snakefile

@@ -0,0 +1,16 @@
+rule dragmap_align:
+    input:
+        reads=["reads/{sample}.1.fastq", "reads/{sample}.2.fastq"],
+        idx="genome",
+    output:
+        "mapped/{sample}.bam",
+    log:
+        "logs/dragmap/{sample}.align.log",
+    params:
+        extra="",
+        sorting="none",  # Can be 'none', 'samtools' or 'picard'.
+        sort_order="queryname",  # Can be 'queryname' or 'coordinate'.
+        sort_extra="",  # Extra args for samtools/picard.
+    threads: 8
+    wrapper:
+        "master/bio/dragmap/align"

bio/dragmap/align/test/Snakefile_picard

@@ -0,0 +1,16 @@
+rule dragmap_align:
+    input:
+        reads=["reads/{sample}.1.fastq", "reads/{sample}.2.fastq"],
+        idx="genome",
+    output:
+        "mapped/{sample}.bam",
+    log:
+        "logs/dragmap/{sample}.align.log",
+    params:
+        extra="",
+        sorting="picard",  # Can be 'none', 'samtools' or 'picard'.
+        sort_order="queryname",  # Can be 'queryname' or 'coordinate'.
+        sort_extra="",  # Extra args for samtools/picard.
+    threads: 8
+    wrapper:
+        "master/bio/dragmap/align"

bio/dragmap/align/test/Snakefile_samtools

@@ -0,0 +1,35 @@
+rule dragmap_align:
+    input:
+        reads=["reads/{sample}.1.fastq", "reads/{sample}.2.fastq"],
+        idx="genome",
+    output:
+        "mapped/{sample}.bam",
+    log:
+        "logs/dragmap/{sample}.align.log",
+    params:
+        extra="",
+        sorting="samtools",  # Can be 'none', 'samtools' or 'picard'.
+        sort_order="queryname",  # Can be 'queryname' or 'coordinate'.
+        sort_extra="",  # Extra args for samtools/picard.
+    threads: 8
+    wrapper:
+        "master/bio/dragmap/align"
+
+
+rule dragmap_align_write_index:
+    input:
+        reads=["reads/{sample}.1.fastq", "reads/{sample}.2.fastq"],
+        idx="genome",
+    output:
+        "mapped_with_index/{sample}.bam",
+        "mapped_with_index/{sample}.bam.csi",
+    log:
+        "logs/dragmap/{sample}.align.log",
+    params:
+        extra="",
+        sorting="samtools",  # Can be 'none', 'samtools' or 'picard'.
+        sort_order="coordinate",  # Can be 'queryname' or 'coordinate'.
+        sort_extra="--write-index",  # Extra args for samtools/picard.
+    threads: 8
+    wrapper:
+        "master/bio/dragmap/align"

bio/dragmap/align/test/genome/hash_table.cfg

@@ -0,0 +1,59 @@
+# Automatically generated by build_hash_table (local version).
+#    Command line: dragen-os --ht-num-threads 2 --build-hash-table true --ht-reference genome.fasta --output-directory genome
+#    Hash table version 8
+#
+# Do not modify.
+
+reference_source     = 'genome.fasta'
+alt_liftover         = ''
+reference_name       = '/home/fgvieira/data/appz/snakemake-wrappers/bio/dragmap/build/test/genome/reference.bin'
+reference_index      = '/home/fgvieira/data/appz/snakemake-wrappers/bio/dragmap/build/test/genome/ref_index.bin'
+reference_sequences  = 1
+reference_len        = 328704
+reference_len_raw    = 20
+reference_len_not_n  = 20
+reference_alt_seed   = 164864
+reference_alt_start  = 164864
+hash_table           = '/home/fgvieira/data/appz/snakemake-wrappers/bio/dragmap/build/test/genome/hash_table.bin'
+extend_table         = '/home/fgvieira/data/appz/snakemake-wrappers/bio/dragmap/build/test/genome/extend_table.bin'
+hash_table_bytes     = 65536
+extend_table_records = 0
+digest_type          = 1
+digest               = 0xC8FF2865
+ref_digest           = 0x00000000
+ref_index_digest     = 0x00000000
+hash_digest          = 0xC8FF2865
+liftover_digest      = 0x00000000
+extend_table_digest  = 0x00000000
+pri_seed_bases       = 17
+max_seed_bases       = 145
+max_ext_increment    = 12
+ref_seed_interval    = 1
+table_addr_bits      = 16
+table_size_64ths     = 64
+max_seed_freq        = 16
+pri_max_seed_freq    = 16
+max_seed_freq_len    = 98
+target_seed_freq     = 4
+min_freq_to_extend   = 17
+thinning_freq_cap    = 12
+max_thinning_factor  = 1
+pri_crc_bits         = 35
+sec_crc_bits         = 35
+seed_len_cost        = 1
+seed_freq_cost       = 0.5
+extension_cost       = 0
+ext_step_cost        = 0.7
+ext_rec_cost         = 4
+repair_strategy      = 0
+min_repair_prob      = 0.2
+anchor_bin_bits      = 0
+hi_freq_rand_hit     = 1
+ext_rand_hit_freq    = 8
+pri_crc_poly         = 666D451CD
+sec_crc_poly         = 666D451CD
+reference_sequence0     = 'Sheila'
+reference_start0        = 163840
+reference_beg_trim0     = 0
+reference_end_trim0     = 0
+reference_len0          = 20

bio/dragmap/align/test/genome/hash_table_stats.txt

@@ -0,0 +1,190 @@
+Reference sequence:
+  Original:         20
+  Encoded:      328704
+  Masked:       328684  (100.0%)
+  Unmasked:         20
+  A bases:          12
+  C bases:           3
+  G bases:           3
+  T bases:           2
+  GC content: 30.0%
+  IUPAC-IUB Codes:
+    0 bases (padding)     :     328684
+    1 base  (A,C,G,T)     :         20
+    2 bases (K,M,R,S,W,Y) :          0
+    3 bases (B,D,H,V)     :          0
+    4 bases (N)           :          0
+
+Reference K-mers: (K=17)
+  Distinct K-mers:              4
+  K-mer positions:              4
+  Palindromes:                  0
+  Total K-mer records:          4
+  Thinned out:                  0
+  Populated seeds:              4
+  NOTE: All K-mer frequency stats are w.r.t. reference K-mer positions,
+        and hence a K-mer with frequency N is included N times.
+  Average K-mer frequency: 1.00
+  K-mer frequency histogram:
+     1   
+     4   
+     100%
+  Log2 K-mer frequency histogram:
+     0   
+     4   
+     100%
+
+Alt contig K-mer positions:           0
+  Liftover K-mer matching:            0  ( 0.0%)
+  Liftover K-mer different:           0  ( 0.0%)
+  No liftover:                        0  ( 0.0%)
+
+Raw primary-seed liftover groups:          0
+  Average liftover group size:          0.00
+  Histogram of liftover group sizes:
+    -
+    -
+    -
+  Histogram of ALT hit count with no liftover:
+    -
+    -
+    -
+
+Liftover groups after possible seed extension:          0
+  Liftover seed matching:              0  ( 0.0%)
+  Liftover seed injected:              0  ( 0.0%)
+  No liftover position:                0  ( 0.0%)
+  Average liftover group size:      0.00
+  Histogram of liftover group sizes:
+    -
+    -
+    -
+  Histogram of ALT hit count with no liftover:
+    -
+    -
+    -
+
+Hash records:
+  Bytes per record:             8
+  Number of records:         8192
+  Hit records:                  4  ( 0.0%)
+  Extension records:            0  ( 0.0%)
+  Interval records:             0  ( 0.0%)
+  Chain records:                0  ( 0.0%)
+  Empty records:             8188  (100.0%)
+  Raw K-mer occupancy:   0.0%
+  Final occupancy:       0.0%
+
+Hash buckets:
+  Records per bucket: 8
+  Number of buckets:  1024
+  Histogram of raw K-mer bucket occupancy:
+     0    1    
+     1020 4    
+     100% 0.39%
+  Histogram of bucket occupancy after extending or rejecting high frequency seeds:
+     0    1    
+     1020 4    
+     100% 0.39%
+  Histogram of physical bucket occupancy as mapped:
+     0    1    
+     1020 4    
+     100% 0.39%
+
+Seed extensions:
+  Base seed length:              17
+  Average extended seed length:  0.0
+  Average extension increment:   0.0
+  Average extension steps:       0.00
+  Extension IDs utilization:     0%
+  Portion of reference K-mers...
+    All raw K-mers:                     4  (100.0%)
+    Extended to longer seeds:           0  (  0.0%)
+    Remaining as primary hit:           4  (100.0%)
+  Space in extension table:             0  (  0.0% of unmasked K-mers)
+  Average frequencies of reference K-mers...
+    All raw K-mers:                  1.00
+    Extended to longer seeds:        0.00
+    Remaining as primary hit:        1.00
+    As extended seed hit:            0.00
+    As primary or extended seed:     1.00
+  Extended seed length histogram:
+    -
+    -
+    -
+  Seed extension increment histogram:
+    -
+    -
+    -
+  Seed extension steps histogram:
+    -
+    -
+    -
+  Pre-extended K-mer frequency histogram:
+    -
+    -
+    -
+  Remaining primary hit K-mer frequency histogram:
+     1   
+     4   
+     100%
+  Post-extended K-mer frequency histogram:
+    -
+    -
+    -
+
+Hash chaining and probing:
+  Number of chains: 0
+  Chain buckets:    0
+  Average length beyond each bucket...
+    chain:  0.0000
+    probe:  0.0000
+    either: 0.0000
+  Histogram of bucket probe lengths replaced by chaining:
+    -
+    -
+    -
+  Bucket chain length histogram:
+     0   
+     1024
+     100%
+  Bucket probe length histogram:
+     0   
+     1024
+     100%
+  Chain or probe length histogram:
+     0   
+     1024
+     100%
+
+Compression:          Records        Bits    Mean
+  auto pri hits:            4          12   3.000
+  auto sec hits:            0         128  128.000
+  auto nul hits:       328700      657400   2.000
+  special hits:             0           0   0.000
+  chain pointers:           0           0   0.000
+  chain ends:               0           0   0.000
+  literals:                 0           0   0.000
+  ext literals:             0           0   0.000
+  TOTAL:               328704      657540   2.000
+  Misc bits:   5595
+  Final bits:  663200
+  Final bytes: 82900
+
+Build thread cycle counts:
+  cyclesOverhead:        6328198
+  cyclesBucketOverhead:  0
+  cyclesExtendPrep:      0
+  cyclesExtendSort:      0
+  cyclesLiftover:        0
+  cyclesPriSort:         0
+  cyclesExtendFreq:      0
+  cyclesExtendDynProg:   0
+  cyclesExtendIntervals: 0
+  cyclesExtendConstruct: 0
+  cyclesBucketSort:      57496
+  cyclesBucketOrganize:  165312
+  cyclesBucketChain:     8960
+  cyclesBucketWrite:     592864
+  cyclesBucketCompress:  53572
+

bio/dragmap/align/test/reads/a.1.fastq

@@ -0,0 +1,4 @@
+@1
+ACGGCAT
++
+!!!!!!!

bio/dragmap/align/test/reads/a.2.fastq

@@ -0,0 +1,4 @@
+@1
+ACGGCAT
++
+!!!!!!!