Skip to content

Latest commit

 

History

History

longread_coverage

README.md

longread_coverage

Galaxy proposed solution

As at February 21, this has been incorporated into the main treevalgal workflow.

Previous progress notes

Partial workflow - missing log coverage and punchlists - TBA.

Partial workflow

Not yet complete. The Galaxy version adds a preconfigured JBrowse browser as an output, and implements the first part of the nearly 300+ lines of DDL needed for the NF TreeVal subworkflow described in the section below. It performs a simple mapping of the reference using minimap for PacBio sample data. Any mapper with bam or other standard output that can be converted into bed would work in place of minimap, for other data.

Sample JBrowse:

(Uses Sanger TinyTestData pac-bio sample, so coverage only seen over a small region):

sample Jbrowse

Zoomed in at a certain point gives a kind of pileup instead of a coverage histogram:

sample Jbrowse zoomed in

The partial workflow is Available in this repository Upload the test reference and the "tiny" PacBio sequence sample. It's too big for github repositories so in Zenodo.

Run the workflow with the correct choice of input files and a Jbrowse history item should eventually appear. Minimap can take a while to run. Warning: The only contig with any sequence coverage in the test sample is ENA|OV656687|OV656687.1 so in Jbrowse, choose that or you won't find any coverage!

Bedtools does coverage so no need for the TreeVal perl script!

Treeval NF DDL subworkflow deconstruction and explanation

Flow chart

Lots of DDL steps Outputs are:

    treeval_upload/
        coverage.bw: Coverage of aligned reads across the reference genome in bigwig format.
        coverage_log.bw: A log corrected coverage file which aims to smooth out the above track.
    treeval_upload/punchlists/
        maxdepth.bigbed: Max read depth punchlist in bigBed format.
        zerodepth.bigbed: Zero read depth punchlist in bigBed format.
        halfcoverage.bigbed: Half read depth punchlist in bigBed format.

The DDL has function calls explained below. Most of the rest of the DDL is not going to be needed other than to figure out exactly how each function gets parameters supplied to the actual command lines.

#!/usr/bin/env nextflow

//
// MODULE IMPORT BLOCK
//
include { BEDTOOLS_BAMTOBED                         } from '../../modules/nf-core/bedtools/bamtobed/main'
include { BEDTOOLS_GENOMECOV                        } from '../../modules/nf-core/bedtools/genomecov/main'
include { BEDTOOLS_MERGE as BEDTOOLS_MERGE_MAX      } from '../../modules/nf-core/bedtools/merge/main'
include { BEDTOOLS_MERGE as BEDTOOLS_MERGE_MIN      } from '../../modules/nf-core/bedtools/merge/main'
include { GNU_SORT                                  } from '../../modules/nf-core/gnu/sort/main'
include { MINIMAP2_INDEX                            } from '../../modules/nf-core/minimap2/index/main'
include { MINIMAP2_ALIGN as MINIMAP2_ALIGN_SPLIT    } from '../../modules/nf-core/minimap2/align/main'
include { MINIMAP2_ALIGN                            } from '../../modules/nf-core/minimap2/align/main'
include { SAMTOOLS_MERGE                            } from '../../modules/nf-core/samtools/merge/main'
include { SAMTOOLS_SORT                             } from '../../modules/nf-core/samtools/sort/main'
include { SAMTOOLS_VIEW                             } from '../../modules/nf-core/samtools/view/main'
include { UCSC_BEDGRAPHTOBIGWIG as BED2BW_NORMAL    } from '../../modules/nf-core/ucsc/bedgraphtobigwig/main'
include { UCSC_BEDGRAPHTOBIGWIG as BED2BW_LOG       } from '../../modules/nf-core/ucsc/bedgraphtobigwig/main'
include { GRAPHOVERALLCOVERAGE                      } from '../../modules/local/graphoverallcoverage'
include { GETMINMAXPUNCHES                          } from '../../modules/local/getminmaxpunches'
include { FINDHALFCOVERAGE                          } from '../../modules/local/findhalfcoverage'
include { LONGREADCOVERAGESCALELOG                  } from '../../modules/local/longreadcoveragescalelog'

workflow LONGREAD_COVERAGE {

    take:
    reference_tuple     // Channel: tuple [ val(meta), file( reference_file ) ]
    dot_genome          // Channel: tuple [ val(meta), [ file( datafile ) ]   ]
    reads_path          // Channel: tuple [ val(meta), val( str )             ]

    main:
    ch_versions             = Channel.empty()

    //
    // MODULE: CREATES INDEX OF REFERENCE FILE
    //
    MINIMAP2_INDEX(
        reference_tuple
    )
    ch_versions             = ch_versions.mix( MINIMAP2_INDEX.out.versions )

    //
    // LOGIC: PREPARE GET_READS_FROM_DIRECTORY INPUT
    //
    reference_tuple
        .combine( reads_path )
        .map { meta, ref, meta2, reads_path ->
            tuple(
                [   id          : meta.id,
                    single_end  : true  ],
                reads_path
            )
        }
        .set { get_reads_input }

    //
    // MODULE: GETS PACBIO READ PATHS FROM READS_PATH
    //
    ch_grabbed_read_paths   = GrabFiles( get_reads_input )

    //
    // LOGIC: PACBIO READS FILES TO CHANNEL
    //
    ch_grabbed_read_paths
        .map { meta, files ->
            tuple( files )
        }
        .flatten()
        .set { ch_read_paths }

    //
    // LOGIC: COMBINE PACBIO READ PATHS WITH MINIMAP2_INDEX OUTPUT
    //
    MINIMAP2_INDEX.out.index
        .combine( ch_read_paths )
        .combine( reference_tuple )
        .map { meta, ref_mmi, read_path, ref_meta, reference ->
            tuple(
                [   id          : meta.id,
                    single_end  : true,
                    split_prefix: read_path.toString().split('/')[-1].split('.fasta.gz')[0]
                ],
                read_path,
                ref_mmi,
                true,
                false,
                false,
                file( reference ).size()
            )
        }
        .branch {
            large               : it[6] > 3000000000
            small               : it[6] < 3000000000
        }
        .set { mma_input }

    mma_input.large
        .multiMap { meta, read_path, ref_mmi, bam_output, cigar_paf, cigar_bam, file_size ->
            read_tuple          : tuple( meta, read_path)
            mmi_index           : ref_mmi
            bool_bam_ouput      : bam_output
            bool_cigar_paf      : cigar_paf
            bool_cigar_bam      : cigar_bam
        }
        .set { large }

    mma_input.small
        .multiMap { meta, read_path, ref_mmi, bam_output, cigar_paf, cigar_bam, file_size ->
            read_tuple          : tuple( meta, read_path)
            mmi_index           : ref_mmi
            bool_bam_ouput      : bam_output
            bool_cigar_paf      : cigar_paf
            bool_cigar_bam      : cigar_bam
        }
        .set { small }

    //
    // MODULE: ALIGN READS TO REFERENCE WHEN REFERENCE <5GB PER SCAFFOLD
    //
    MINIMAP2_ALIGN (
        small.read_tuple,
        small.mmi_index,
        small.bool_bam_ouput,
        small.bool_cigar_paf,
        small.bool_cigar_bam
    )
    ch_versions             = ch_versions.mix(MINIMAP2_ALIGN.out.versions)
    ch_align_bams           = MINIMAP2_ALIGN.out.bam

    //
    // MODULE: ALIGN READS TO REFERENCE WHEN REFERENCE >5GB PER SCAFFOLD
    //
    MINIMAP2_ALIGN_SPLIT (
        large.read_tuple,
        large.mmi_index,
        large.bool_bam_ouput,
        large.bool_cigar_paf,
        large.bool_cigar_bam
    )
    ch_versions             = ch_versions.mix(MINIMAP2_ALIGN_SPLIT.out.versions)

    //
    // LOGIC: COLLECT OUTPUTTED BAM FILES FROM BOTH PROCESSES
    //
    ch_align_bams
        .mix( MINIMAP2_ALIGN_SPLIT.out.bam )
        .set { ch_bams }

    //
    // LOGIC: PREPARING MERGE INPUT WITH REFERENCE GENOME AND REFERENCE INDEX
    //
    ch_bams
        .map { meta, file ->
            tuple( file )
        }
        .collect()
        .map { file ->
            tuple (
                [ id    : file[0].toString().split('/')[-1].split('_')[0] ], // Change sample ID
                file
            )
        }
        .set { collected_files_for_merge }

    //
    // MODULE: MERGES THE BAM FILES IN REGARDS TO THE REFERENCE
    //         EMITS A MERGED BAM
    SAMTOOLS_MERGE(
        collected_files_for_merge,
        reference_tuple,
        [[],[]]
    )
    ch_versions             = ch_versions.mix(SAMTOOLS_MERGE.out.versions)

    //
    // MODULE: SORT THE MERGED BAM BEFORE CONVERSION
    //
    SAMTOOLS_SORT (
        SAMTOOLS_MERGE.out.bam
    )
    ch_versions             = ch_versions.mix( SAMTOOLS_MERGE.out.versions )

    //
    // LOGIC: PREPARING MERGE INPUT WITH REFERENCE GENOME AND REFERENCE INDEX
    //
    SAMTOOLS_SORT.out.bam
        .combine( reference_tuple )
        .multiMap { meta, bam, ref_meta, ref ->
                bam_input       :   tuple(
                                        [   id          : meta.id,
                                            sz          : bam.size(),
                                            single_end  : true  ],
                                        bam,
                                        []   // As we aren't using an index file here
                                    )
                ref_input       :   tuple(
                                        ref_meta,
                                        ref
                                    )
        }
        .set { view_input }
    //
    // MODULE: EXTRACT READS FOR PRIMARY ASSEMBLY
    //
    SAMTOOLS_VIEW(
        view_input.bam_input,
        view_input.ref_input,
        []
    )
    ch_versions             = ch_versions.mix(SAMTOOLS_VIEW.out.versions)

    //
    // MODULE: BAM TO PRIMARY BED
    //
    BEDTOOLS_BAMTOBED(
        SAMTOOLS_VIEW.out.bam
    )
    ch_versions             = ch_versions.mix(BEDTOOLS_BAMTOBED.out.versions)

    //
    // LOGIC: PREPARING Genome2Cov INPUT
    //
    BEDTOOLS_BAMTOBED.out.bed
        .combine( dot_genome )
        .multiMap { meta, file, my_genome_meta, my_genome ->
            input_tuple         :   tuple (
                                        [   id          :   meta.id,
                                            single_end  :   true    ],
                                        file,
                                        1
                                    )
            dot_genome          :   my_genome
            file_suffix         :   'bed'
        }
        .set { genomecov_input }

    //
    // MODULE: Genome2Cov
    //
    BEDTOOLS_GENOMECOV(
        genomecov_input.input_tuple,
        genomecov_input.dot_genome,
        genomecov_input.file_suffix
    )
    ch_versions             = ch_versions.mix(BEDTOOLS_GENOMECOV.out.versions)

    //
    // MODULE: SORT THE PRIMARY BED FILE
    //
    GNU_SORT(
        BEDTOOLS_GENOMECOV.out.genomecov
    )
    ch_versions             = ch_versions.mix(GNU_SORT.out.versions)

    //
    // MODULE: get_minmax_punches
    //
    GETMINMAXPUNCHES(
        GNU_SORT.out.sorted
    )
    ch_versions             = ch_versions.mix(GETMINMAXPUNCHES.out.versions)

    //
    // MODULE: get_minmax_punches
    //
    BEDTOOLS_MERGE_MAX(
        GETMINMAXPUNCHES.out.max
    )
    ch_versions             = ch_versions.mix(BEDTOOLS_MERGE_MAX.out.versions)

    //
    // MODULE: get_minmax_punches
    //
    BEDTOOLS_MERGE_MIN(
        GETMINMAXPUNCHES.out.min
    )
    ch_versions             = ch_versions.mix(BEDTOOLS_MERGE_MIN.out.versions)

    //
    // MODULE: GENERATE DEPTHGRAPH
    //
    GRAPHOVERALLCOVERAGE(
        GNU_SORT.out.sorted
    )
    ch_versions             = ch_versions.mix(GRAPHOVERALLCOVERAGE.out.versions)
    ch_depthgraph           = GRAPHOVERALLCOVERAGE.out.part

    //
    // LOGIC: PREPARING FINDHALFCOVERAGE INPUT
    //
    GNU_SORT.out.sorted
        .combine( GRAPHOVERALLCOVERAGE.out.part )
        .combine( dot_genome )
        .multiMap { meta, file, meta_depthgraph, depthgraph, meta_my_genome, my_genome ->
            halfcov_bed     :       tuple( [ id : meta.id, single_end : true  ], file )
            genome_file     :       my_genome
            depthgraph_file :       depthgraph
        }
        .set { halfcov_input }

    //
    // MODULE: FIND REGIONS OF HALF COVERAGE
    //
    FINDHALFCOVERAGE(
        halfcov_input.halfcov_bed,
        halfcov_input.genome_file,
        halfcov_input.depthgraph_file
    )
    ch_versions             = ch_versions.mix(FINDHALFCOVERAGE.out.versions)

    //
    // LOGIC: PREPARING NORMAL COVERAGE INPUT
    //
    GNU_SORT.out.sorted
        .combine( dot_genome )
        .combine(reference_tuple)
        .multiMap { meta, file, meta_my_genome, my_genome, ref_meta, ref ->
            ch_coverage_bed :   tuple ([ id: ref_meta.id, single_end: true], file)
            genome_file     :   my_genome
        }
        .set { bed2bw_normal_input }

    //
    // MODULE: CONVERT BEDGRAPH TO BIGWIG FOR NORMAL COVERAGE
    //
    BED2BW_NORMAL(
        bed2bw_normal_input.ch_coverage_bed,
        bed2bw_normal_input.genome_file
    )
    ch_versions             = ch_versions.mix(BED2BW_NORMAL.out.versions)

    //
    // MODULE: CONVERT COVERAGE TO LOG
    //
    LONGREADCOVERAGESCALELOG(
        GNU_SORT.out.sorted
    )
    ch_versions             = ch_versions.mix(LONGREADCOVERAGESCALELOG.out.versions)

    //
    // LOGIC: PREPARING LOG COVERAGE INPUT
    //
    LONGREADCOVERAGESCALELOG.out.bed
        .combine( dot_genome )
        .combine(reference_tuple)
        .multiMap { meta, file, meta_my_genome, my_genome, ref_meta, ref ->
            ch_coverage_bed :   tuple ([ id: ref_meta.id, single_end: true], file)
            genome_file     :   my_genome
        }
        .set { bed2bw_log_input }

    //
    // MODULE: CONVERT BEDGRAPH TO BIGWIG FOR LOG COVERAGE
    //
    BED2BW_LOG(
        bed2bw_log_input.ch_coverage_bed,
        bed2bw_log_input.genome_file
    )
    ch_versions             = ch_versions.mix(BED2BW_LOG.out.versions)

    //
    // LOGIC: GENERATE A SUMMARY TUPLE FOR OUTPUT
    //
    ch_grabbed_read_paths
            .collect()
            .map { meta, fasta ->
                tuple( [    id: 'pacbio',
                            sz: fasta instanceof ArrayList ? fasta.collect { it.size()} : fasta.size() ],
                            fasta
                )
            }
            .set { ch_reporting_pacbio }

    emit:
    ch_minbed               = BEDTOOLS_MERGE_MIN.out.bed
    ch_halfbed              = FINDHALFCOVERAGE.out.bed
    ch_maxbed               = BEDTOOLS_MERGE_MAX.out.bed
    ch_reporting            = ch_reporting_pacbio.collect()
    ch_covbw_nor            = BED2BW_NORMAL.out.bigwig
    ch_covbw_log            = BED2BW_LOG.out.bigwig
    versions                = ch_versions
}

process GrabFiles {
    label 'process_tiny'

    tag "${meta.id}"
    executor 'local'

    input:
    tuple val(meta), path("in")

    output:
    tuple val(meta), path("in/*.fasta.gz")

    "true"
}

[MINIMAP2_INDEX](https://github.com/sanger-tol/treeval/blob/dev/modules/nf-core/minimap2/index/main.nf) uses *conda "bioconda::minimap2=2.24"* to execute

minimap2 \ -t $task.cpus \ -d ${fasta.baseName}.mmi \ $args \ $fasta

There is some DDL to split very large inputs before calling [MINIMAP2_ALIGN](https://github.com/sanger-tol/treeval/blob/dev/modules/nf-core/minimap2/align/main.nf)

// // MODULE: ALIGN READS TO REFERENCE WHEN REFERENCE <5GB PER SCAFFOLD //


that uses *conda "bioconda::minimap2=2.24 bioconda::samtools=1.14"*. It is called with "small" inputs to execute:

minimap2 \ $args \ -t $task.cpus \ "${reference ?: reads}" \ "$reads" \ $cigar_paf \ $set_cigar_bam \ $bam_output

then the same MINIMAP2 function renamed as MINIMAP2_ALIGN_SPLIT is called with the large ones. No idea why this renamed function and logic to
adjustments ids and merge - there is a comment:

// // MODULE: MERGES THE BAM FILES IN REGARDS TO THE REFERENCE // EMITS A MERGED BAM


using [SAMTOOLS_MERGE](https://github.com/sanger-tol/treeval/blob/dev/modules/nf-core/samtools/merge/main.nf) is called - also in [nuc_alignments](nuc_alignments) and [hic_mapping](hic_mapping)
uses *conda "bioconda::samtools=1.17"* to run

samtools \ merge \ --threads ${task.cpus-1} \ $args \ ${reference} \ ${prefix}.${file_type} \ $input_files

[SAMTOOLS_SORT](https://github.com/sanger-tol/treeval/blob/dev/modules/nf-core/samtools/sort/main.nf) follows using *conda "bioconda::samtools=1.17"* to execute:

samtools sort \ $args \ -@ $task.cpus \ -o ${prefix}.bam \ -T $prefix \ $bam


[SAMTOOLS_VIEW](https://github.com/sanger-tol/treeval/blob/dev/modules/nf-core/samtools/view/main.nf) runs:

samtools \ view \ --threads ${task.cpus-1} \ ${reference} \ ${readnames} \ $args \ -o ${prefix}.${file_type} \ $input \ $args2


[BEDTOOLS_BAMTOBED](https://github.com/sanger-tol/treeval/blob/dev/modules/nf-core/bedtools/bamtobed/main.nf) uses *conda "bioconda::bedtools=2.31.0"* to run:

bedtools \ bamtobed \ $args \ -i $bam \ &gt; ${prefix}.bed


[BEDTOOLS_GENOMECOV](https://github.com/sanger-tol/treeval/blob/dev/modules/nf-core/bedtools/genomecov/main.nf) uses *bioconda::bedtools=2.31.0"*.
That runs one of two separate command lines depending on whether the inputs are .bam files using odd DDL logic:

if (intervals.name =~ /.bam/) { """ bedtools \ genomecov \ -ibam $intervals \ $args \ &gt; ${prefix}.${extension}

    cat <<-END_VERSIONS > versions.yml
    "${task.process}":
        bedtools: \$(bedtools --version | sed -e "s/bedtools v//g")
    END_VERSIONS
    """
} else {
    """
    bedtools \\
        genomecov \\
        -i $intervals \\
        -g $sizes \\
        $args \\
        > ${prefix}.${extension}

[GNU_SORT](https://github.com/sanger-tol/treeval/blob/dev/modules/nf-core/gnu/sort/main.nf) uses coreutils sort on the output merge called as:

sort ${args} ${input} > ${output_file}


The sorted file is passed to [GETMINMAXPUNCHES](https://github.com/sanger-tol/treeval/blob/dev/modules/local/getminmaxpunches.nf) to run awk:

$/ cat "${bedfile}"
| awk '{ if ($4 == 0) {print $0 >> "zero.bed" } else if ($4 > 1000) {print $0 >> "max.bed"}}'

BEDTOOLS_MERGE_MIN and MAX are aliases for [BEDTOOLS_MERGE](https://github.com/sanger-tol/treeval/blob/dev/modules/nf-core/bedtools/merge/main.nf) using *"bioconda::bedtools=2.31.0"* to execute:

bedtools \ merge \ -i $bed \ $args \ &gt; ${prefix}.bed

It is called separately on the min/max output files from the GETMINMAXPUNCHES step above.

A depth graph is generated using [GRAPHOVERALLCOVERAGE](https://github.com/sanger-tol/treeval/blob/dev/modules/local/graphoverallcoverage.nf) using *"conda-forge::perl=5.26.2"* to run
a perl script from the tree/bin directory

graph_overall_coverage.pl $bed &gt; ${prefix}.part


The script is:

#!/usr/bin/env perl

Script originally developed by Yumi Sims (yy5@sanger.ac.uk)

use warnings;

my $file = shift;

my ($file) = @ARGV;

if (!@ARGV || ($ARGV[0] eq '--version')) { print "1.0\n"; exit 0; }

open (FILE, $file) || die "can't open file $file\n";

my %depthcount; while (my $line = ) { chomp $line; my ($id, $start, $end, $depth) = split ("\t", $line); my $length = $end - $start;

if ($depthcount{$depth}){
    $depthcount{$depth} += $length;
}
else {
    $depthcount{$depth} = $length;
}

}

foreach my $depth (sort {$a<=>$b} keys %depthcount){ print join("\t", $depth, $depthcount{$depth}) ."\n"; }

[FINDHALFCOVERAGE](https://github.com/sanger-tol/treeval/blob/dev/modules/local/findhalfcoverage.nf) runs a [python script](https://github.com/sanger-tol/treeval/blob/dev/bin/findHalfcoverage.py) from the bin directory
using the command:

findHalfcoverage.py -c $bedfile -m $my_genome -d $depthgraph &gt; ${prefix}.bed


GNU_SORT is used again, followed by BED2BW_NORMAL - that's an alias for [ucsc_bedgraphtobigwig](https://github.com/sanger-tol/treeval/blob/dev/modules/nf-core/ucsc/bedgraphtobigwig/main.nf) that calls


bedGraphToBigWig \\
        $bedgraph \\
        $sizes \\
        ${prefix}.bigWig


There is an existing bedgraph2BigWig wrapper: https://usegalaxy.eu/root?tool_id=wig_to_bigWig

Then there is a repeat of the above steps after log scaling with

[LONGREADCOVERAGESCALELOG](https://github.com/sanger-tol/treeval/blob/dev/modules/local/longreadcoveragescalelog.nf) that runs a python script from the /tree/bin directory using

BG: This script can be replaced by https://usegalaxy.eu/root?tool_id=toolshed.g2.bx.psu.edu/repos/devteam/column_maker/Add_a_column1/2.0

longread_cov_log.py -i $bedfile &gt; ${prefix}.bed

That python script is

#!/usr/bin/env python

import optparse import math

Script originally developed by Will Eagles (we3@sanger.ac.uk)

def process_line(line): line_values = line.rsplit(None, 1)

try:
    cov_val = float(line_values[1])
except:
    cov_val = 0

if cov_val > 0:
    log_cov_val = math.log(cov_val)
else:
    log_cov_val = 0

return line_values[0] + "\t" + str(round(log_cov_val, 2))

def main(): parser = optparse.OptionParser(version="%prog 1.0") parser.add_option( "-i", "--inputfile", dest="inputfile", default="default.input", )

options, remainder = parser.parse_args()

cov_bed = open(options.inputfile, "r")

for line in cov_bed:
    print(process_line(line))

if name == "main": main()


The output is processed with BED2BW_log, another alias for [ucsc_bedgraphtobigwig](https://github.com/sanger-tol/treeval/blob/dev/modules/nf-core/ucsc/bedgraphtobigwig/main.nf) that calls


bedGraphToBigWig \\
        $bedgraph \\
        $sizes \\
        ${prefix}.bigWig