add bowtie1

Juke34 · Jan 16, 2025 · fe015d7 · fe015d7
1 parent c7343c3
commit fe015d7
Show file tree

Hide file tree

Showing 8 changed files with 109 additions and 10 deletions.
diff --git a/README.md b/README.md
@@ -39,6 +39,7 @@ You can choose to run one or several aligner in parallel.
 | Tool	| Single End (short reads) | Paired end (short reads) | Pacbio | ONT |
 | --- | --- | --- |  --- | --- |
 | bbmap | ✅ | ✅ | ⚠️ | ⚠️ |
+| bowtie | ✅ | ✅ | ⚠️ | ⚠️ |
 | bowtie2 | ✅ | ✅ | ⚠️ | ⚠️ |
 | bwaaln | ✅ | ✅ R1 and R2 independently aligned then merged with bwa sampe | ✅ | ✅ |
 | bwamem | ✅ | ✅ | ⚠️ | ⚠️ |
@@ -72,6 +73,7 @@ It is then translated to the correct option in the following aligners:
 | --- | --- | --- | --- |
 | bbmap | xs=fr / xs=ss / xs=us | ISF ISR / OSF OSR / U | strand information |
 | bbmap | - / rcs=f / | ISF ISR IU / OSF OSR OU MSF MSR MU | read orientation |
+| bowtie | --fr / --rf / --ff |  ISF ISR IU / OSF OSR OU / MSF MSR MU| read orientation |
 | bowtie2 | --fr / --rf / --ff |  ISF ISR IU / OSF OSR OU / MSF MSR MU| read orientation |
 | bwaaln | 🚫 | 🚫 | 🚫 |
 | bwamem | 🚫 | 🚫 | 🚫 |
@@ -104,8 +106,9 @@ If the `skip_libray_usage` paramater is set, the information about the library t
 If you provide an annotation file the pipeline will pass automatically the file to the following aligner:  
 
 | Tool	| accept | 
-| --- | --- | 
-| bbmap | 🚫 | 
+| --- | --- |
+| bbmap | 🚫 |
+| bowtie | 🚫 |
 | bowtie2 | 🚫 |
 | bwaaln | 🚫 |
 | bwamem | 🚫 |

diff --git a/aline.nf b/aline.nf
@@ -38,9 +38,10 @@ params.annotation = ""
 params.trimming_fastp = false
 
 // Aligner params
-align_tools = [ 'bbmap', 'bowtie2', 'bwaaln', 'bwamem', 'bwasw', 'graphmap2', 'hisat2', 'kallisto', 'minimap2', 'novoalign', 'nucmer', 'ngmlr', 'star', 'subread', 'sublong', 'tophat2' ]
+align_tools = [ 'bbmap', 'bowtie', 'bowtie2', 'bwaaln', 'bwamem', 'bwasw', 'graphmap2', 'hisat2', 'kallisto', 'minimap2', 'novoalign', 'nucmer', 'ngmlr', 'star', 'subread', 'sublong', 'tophat2' ]
 params.aligner = ''
 params.bbmap_options = ''
+params.bowtie_options = ''
 params.bowtie2_options = ''
 params.bwaaln_options = ''
 params.bwamem_options = ''
@@ -358,12 +359,13 @@ log.info printAlignerOptions(aligner_list, annotation_file, params.star_index_op
 //*************************************************
 include {read_length; set_tuple_withUserReadLength} from "$baseDir/modules/bash.nf" 
 include {bbmap_index; bbmap} from "$baseDir/modules/bbmap.nf"
+include {bowtie_index; bowtie} from "$baseDir/modules/bowtie.nf"
 include {bowtie2_index; bowtie2} from "$baseDir/modules/bowtie2.nf"
 include {bwa_index; bwaaln; bwamem; bwasw} from "$baseDir/modules/bwa.nf"
 include {seqkit_convert} from "$baseDir/modules/seqkit.nf"
 include {graphmap2_index; graphmap2} from "$baseDir/modules/graphmap2.nf"
 include {fastp} from "$baseDir/modules/fastp.nf"
-include {fastqc as fastqc_raw; fastqc as fastqc_fastp; fastqc as fastqc_ali_bbmap; fastqc as fastqc_ali_bowtie2 ; 
+include {fastqc as fastqc_raw; fastqc as fastqc_fastp; fastqc as fastqc_ali_bbmap; fastqc as fastqc_ali_bowtie ; fastqc as fastqc_ali_bowtie2 ; 
          fastqc as fastqc_ali_bwaaln; fastqc as fastqc_ali_bwamem; fastqc as fastqc_ali_bwasw; fastqc as fastqc_ali_graphmap2 ; 
          fastqc as fastqc_ali_hisat2; fastqc as fastqc_ali_kallisto; fastqc as fastqc_ali_minimap2; fastqc as fastqc_ali_ngmlr; 
          fastqc as fastqc_ali_novoalign ; fastqc as fastqc_ali_nucmer; fastqc as fastqc_ali_star; fastqc as fastqc_ali_subread ; 
@@ -376,11 +378,11 @@ include {ngmlr} from "$baseDir/modules/ngmlr.nf"
 include {nucmer} from "$baseDir/modules/mummer4.nf" 
 include {novoalign_index; novoalign} from "$baseDir/modules/novoalign.nf"
 include {salmon_index; salmon_guess_lib; set_tuple_withUserLib} from "$baseDir/modules/salmon.nf" 
-include {samtools_sam2bam_nucmer; samtools_sam2bam as samtools_sam2bam_bowtie2; samtools_sam2bam as samtools_sam2bam_bwaaln; 
+include {samtools_sam2bam_nucmer; samtools_sam2bam as samtools_sam2bam_bowtie; samtools_sam2bam as samtools_sam2bam_bowtie2; samtools_sam2bam as samtools_sam2bam_bwaaln; 
          samtools_sam2bam as samtools_sam2bam_bwamem; samtools_sam2bam as samtools_sam2bam_bwasw; samtools_sam2bam as samtools_sam2bam_graphmap2; 
          samtools_sam2bam as samtools_sam2bam_hisat2; samtools_sam2bam as samtools_sam2bam_minimap2; 
          samtools_sam2bam as samtools_sam2bam_ngmlr; samtools_sam2bam as samtools_sam2bam_novoalign } from "$baseDir/modules/samtools.nf"
-include {samtools_sort as samtools_sort_bbmap; samtools_sort as samtools_sort_bowtie2; samtools_sort as samtools_sort_bwaaln; 
+include {samtools_sort as samtools_sort_bbmap; samtools_sort as samtools_sort_bowtie; samtools_sort as samtools_sort_bowtie2; samtools_sort as samtools_sort_bwaaln; 
          samtools_sort as samtools_sort_bwamem; samtools_sort as samtools_sort_bwasw; samtools_sort as samtools_sort_graphmap2; 
          samtools_sort as samtools_sort_hisat2; samtools_sort as samtools_sort_minimap2; samtools_sort as samtools_sort_ngmlr; 
          samtools_sort as samtools_sort_novoalign;  samtools_sort as samtools_sort_nucmer; samtools_sort as samtools_sort_tophat2;
@@ -614,6 +616,22 @@ workflow align {
             }
         }
 
+        // ------------------- BOWTIE -----------------
+        if ( "bowtie" in aligner_list ){ // &&
+            bowtie_index(genome.collect(), "alignment/bowtie/indicies") // index
+            bowtie(reads, genome.collect(), bowtie_index.out.collect(), "alignment/bowtie") // align
+            logs.concat(bowtie.out.bowtie_summary).set{logs} // save log
+            // convert sam to bam
+            samtools_sam2bam_bowtie(bowtie.out.tuple_sample_sam)
+            // sort
+            samtools_sort_bowtie(samtools_sam2bam_bowtie.out.tuple_sample_bam, "alignment/bowtie")
+            // stat on aligned reads
+            if(params.fastqc){
+                fastqc_ali_bowtie(samtools_sort_bowtie.out.tuple_sample_sortedbam, "fastqc/bowtie", "bowtie")
+                logs.concat(fastqc_ali_bowtie.out).set{logs} // save log
+            }
+        }
+
         // ------------------- BOWTIE2 -----------------
         if ( "bowtie2" in aligner_list ){ // &&
             bowtie2_index(genome.collect(), "alignment/bowtie2/indicies") // index
@@ -911,6 +929,7 @@ def helpMSG() {
 
     Aligner specific options
         --bbmap_options             additional options for bbmap
+        --bowtie_options            additional options for bowtie
         --bowtie2_options           additional options for bowtie2
         --bwaaln_options            additional options for bwaaln
         --bwamem_options            additional options for bwamem
@@ -943,6 +962,11 @@ def printAlignerOptions(aligner_list, annotation_file, star_index_options) {
         bbmap_tool                 : ${bbmap_tool}
         bbmap_options              : ${params.bbmap_options}
     """} 
+    if ("bowtie" in aligner_list){ 
+        sentence += """       
+    bowtie parameters
+        bowtie_options            : ${params.bowtie_options}
+    """} 
     if ("bowtie2" in aligner_list){ 
         sentence += """       
     bowtie2 parameters

diff --git a/config/softwares.config b/config/softwares.config
@@ -5,6 +5,9 @@ process {
     withLabel: 'bbmap' {
         container = 'quay.io/biocontainers/bbmap:39.08--h92535d8_0'
     }
+    withLabel: 'bowtie' {
+        container = 'quay.io/biocontainers/bowtie:1.3.1--py39h9046dc2_10'
+  	}
     withLabel: 'bowtie2' {
         container = 'quay.io/biocontainers/bowtie2:2.5.1--py38he00c5e5_2'
   	}

diff --git a/modules/bowtie.nf b/modules/bowtie.nf
@@ -0,0 +1,69 @@
+process bowtie_index {
+    label 'bowtie'
+    tag "$genome_fasta"
+    publishDir "${params.outdir}/${outpath}", mode: 'copy'
+
+    input:
+        path(genome_fasta)
+        val outpath
+
+    output:
+        path('*.ebwt')
+
+    script:
+
+        """
+        bowtie-build --threads ${task.cpus} $genome_fasta ${genome_fasta.baseName}
+        """
+}
+
+process bowtie {
+    label 'bowtie'
+    tag "$sample"
+    publishDir "${params.outdir}/${outpath}", pattern: "*bowtie.log", mode: 'copy'
+
+    input:
+        tuple val(sample), path(reads), val(library), val(read_length)
+        path genome
+        path index_files
+        val outpath
+
+    output:
+        tuple val(sample), path ("*.sam"), emit: tuple_sample_sam
+        path "*bowtie.log",  emit: bowtie_summary
+
+    script:
+
+        def read_orientation=""
+        if (! params.bowtie_options.contains("--fr ") && 
+            ! params.bowtie_options.contains("--rf ") && 
+            ! params.bowtie_options.contains("--ff ") &&
+            params.read_type == "short_paired" && 
+            ! params.skip_libray_usage){ 
+            if (library.contains("I") ){
+                read_orientation = "--fr"
+            } else if (library.contains("O") ){
+                read_orientation = "--rf"
+            } else if (library.contains("M") ){
+                read_orientation = "--ff"
+            }  
+        }
+
+        if (params.read_type == "short_paired"){
+        """
+            bowtie ${params.bowtie_options} ${read_orientation}\\
+                -p ${task.cpus} \\
+                -x ${genome.baseName} \\
+                -S ${reads[0].baseName.replace('.fastq','')}_bowtie.sam \\
+                -1 ${reads[0]} -2 ${reads[1]}  2> ${reads[0].baseName.replace('.fastq','')}_bowtie.log
+        """
+        } else {
+        """
+            bowtie ${params.bowtie_options} ${read_orientation}\\
+                    -p ${task.cpus} \\
+                    -x ${genome.baseName} \\
+                    -S ${reads} > ${reads.baseName.replace('.fastq','')}_bowtie.sam 2> ${reads.baseName}_bowtie.log
+        """
+        }
+
+}
diff --git a/profiles/test_illumina_paired.config b/profiles/test_illumina_paired.config
@@ -7,7 +7,7 @@
 params {
     reads = "$baseDir/test/illumina/"
     genome = "$baseDir/test/yeast.fa"
-    aligner = 'bbmap,bowtie2,bwaaln,bwamem,bwasw,graphmap2,hisat2,minimap2,nucmer,star,subread,tophat2'
+    aligner = 'bbmap,bowtie,bowtie2,bwaaln,bwamem,bwasw,graphmap2,hisat2,minimap2,nucmer,star,subread,tophat2'
     star_options = "--genomeSAindexNbases 9" // the default 14 is too large for the genome size=1351857
     multiqc_config = "$baseDir/config/multiqc_conf.yml"
 }
diff --git a/profiles/test_illumina_single.config b/profiles/test_illumina_single.config
@@ -8,7 +8,7 @@ params {
     reads = "$baseDir/test/illumina/"
     genome = "$baseDir/test/yeast.fa"
     params.read_type = "short_single"
-    aligner = 'bbmap,bowtie2,bwaaln,bwamem,bwasw,graphmap2,hisat2,kallisto,minimap2,ngmlr,nucmer,star,subread,sublong,tophat2'
+    aligner = 'bbmap,bowtie,bowtie2,bwaaln,bwamem,bwasw,graphmap2,hisat2,kallisto,minimap2,ngmlr,nucmer,star,subread,sublong,tophat2'
     trimming_fastp = true
     star_options = "--genomeSAindexNbases 9" // the default 14 is too large for the genome size=1351857
     multiqc_config = "$baseDir/config/multiqc_conf.yml"

diff --git a/profiles/test_ont.config b/profiles/test_ont.config
@@ -8,7 +8,7 @@ params {
     reads = "$baseDir/test/nanopore"
     genome = "$baseDir/test/yeast.fa"
     read_type = "ont"
-    aligner = 'bwaaln,bwamem,bwasw,graphmap2,hisat2,kallisto,minimap2,ngmlr,nucmer,star,subread,sublong'
+    aligner = 'bbmap,bowtie,bowtie2,bwaaln,bwamem,bwasw,graphmap2,hisat2,kallisto,minimap2,ngmlr,nucmer,star,subread,sublong'
     library_type = 'U'
     star_options = '--outFilterMismatchNmax 100 --seedSearchLmax 30   --seedSearchStartLmax 30 --seedPerReadNmax 100000   --seedPerWindowNmax 100 --alignTranscriptsPerReadNmax 100000 --alignTranscriptsPerWindowNmax 10000'
     multiqc_config = "$baseDir/config/multiqc_conf.yml"

diff --git a/profiles/test_pacbio.config b/profiles/test_pacbio.config
@@ -8,7 +8,7 @@ params {
     reads = "$baseDir/test/pacbio/"
     genome = "$baseDir/test/yeast.fa"
     read_type = "pacbio"
-    aligner = 'bwaaln,bwamem,bwasw,graphmap2,hisat2,kallisto,minimap2,ngmlr,nucmer,star,subread,sublong'
+    aligner = 'bbmap,bowtie,bowtie2,bwaaln,bwamem,bwasw,graphmap2,hisat2,kallisto,minimap2,ngmlr,nucmer,star,subread,sublong'
     star_options = '--outFilterMismatchNmax 100 --seedSearchLmax 30   --seedSearchStartLmax 30 --seedPerReadNmax 100000 --seedPerWindowNmax 100 --alignTranscriptsPerReadNmax 100000 --alignTranscriptsPerWindowNmax 10000'
     star_index_options = '--genomeSAindexNbases 9'
     multiqc_config = "$baseDir/config/multiqc_conf.yml"