Skip to content

Latest commit

 

History

History
146 lines (88 loc) · 4.38 KB

README.md

File metadata and controls

146 lines (88 loc) · 4.38 KB
Raw

Assemblatron - De novo assembly tool kit

Assemblatron is a toolkit for De novo assembly and variant calling of de novo assemblies. type python assemblatron.py --help

For a help message of the avialable tools.

The workflow

The full workflow involves:

1: assembly using Fermi2 (similar to workflow as fermiKit)

2: alignment of the contigs

3: Quality control

4: variant calling

However, you may start or end the analysis as any step. The assemblatron workflow is run through the following commands:

Install

run the install script: ./INSTALL.sh The install script will download and install fermikit,kmc, and bfc-kmc. Dependencies:

conda
samtools
python 2.7
bwa
numpy
minimap2
sambamba

Assemblatron may also be run using the singularity container:

Download the container:

singularity pull --name assemblatron.img shub://J35P312/Assemblatron

Run assemblatron through the singularity container:

singularity exec assemblatron.img python /bin/assemblatron.py --help

Assemble

Assemblatron performs de novo assembly using a pipeline similar to fermikit. The pipeline is run through the following commands:

	python assemblatron.py --assemble --fastq <input.fastq> --prefix <prefix>

You may also align the contigs to the reference genome:

	python assemblatron.py --assemble --ref <reference.fasta> --fastq <input.fastq> --prefix <prefix> --align

Per default, the Assembly module will generate an output file named .fastq. If asssemblatron was run with the --align option, a bam file named .bam will be produced as well.

Type help for more info:

	 python assemblatron.py --assemble --help

The main differences between fermikit and Assemblatron is that Assemblatron uses the kmc branch of bfc. Additionally, Assemblatron uses piping more extensively, hence it will not create any intermidiate fastq files. For more info on Fermikit, visit the Fermikit website:

	https://github.com/lh3/fermikit

Align

Assemblatron performs alignment using bwa mem. The output is printed to a file named <prefix.bam>. The input is asssumed to be contigs produced through any assembly process.

python assemblatron.py --align --ref <reference.fasta> --contigs <contigs.fasta> --prefix <prefix>

A bam file named .bam will be produced

Stats

compute various statistics of an assembly. The output is printed to stdout.

python assemblatron.py --stats --bam <contigs_bam>

The statistics include N50, L50, assembly size, and the number of contigs.

The statistics module supports any number of fasta files. Type --help for more information. NOTE: use absolute path for the output directory.

SV

Call SV. Assemblatron will classify variants as DEL, INV, BND (translocation or complex), INS, and TDUP.

python assemblatron.py --sv --bam <contigs.bam> > out.vcf

The output is printed to stdout. Note, the SV-caller assumes that the base qualitiy scores represent the base coverage.

other options:

-h, --help            show this help message and exit
--sv                  call SV from the aligned contigs
--bam BAM             input bam (contigs
--q Q                 minimum allowed mapping quality(default = 10)
--len_ctg LEN_CTG     minimum uniquely mapped contig length (default = 40)
--max_coverage MAX_COVERAGE
--sample              sample id (default = bam-filename)
                    calls from regions exceeding the maximum coverage are
                    filtered
--min_size MIN_SIZE   minimum variant size (default=100)

SNV

Call indels and SNV using htsbox pileup.

the output is printed to stdout

options:

-h, --help show this help message and exit --bam BAM input bam (contigs) --ref REF reference fasta

fastq

convert a bam file to fastq.

python assemblatron.py --fastq <input.bam>

The output is printed to stdout.

fasta

Convert a contig bam file to fasta.

python assemblatron.py --fastq <contigs.bam>

The output is printed to stdout.

Testing

De novo assembly of NA12878 may be downloaded from the 10X website:

Start by aligning the contigs using the align module, thereafter you may call SV and SNV using the sv and snv modules. Truthsets are available at the GIAB website:

As well as through the SV classify supplementary methods:

Cite

Cite the components that you used, as well as the Assemblatron git hub page.

If you performed De novo assembly, cite the FermiKit paper:

	https://github.com/lh3/fermikit