ContigExtender, was developed to extend contigs, complementing de novo assembly. ContigExtender employs a novel recursive Overlap Layout Candidates (r-OLC) strategy that explores multiple extending paths to achieve longer and highly accurate contigs. ContigExtender is effective for extending contigs significantly in in silico synthesized and real metagenomics datasets.
The only supported operating system is Linux.
To install, first install Docker
Then, clone this repository with
git clone https://github.com/dengzac/contig-extender.git
cd contig-extender
Finally, run ./dist/extender_wrapper. This script will automatically download the Docker image and run ContigExtender with the parameters that are passed to it.
The examples folder contains a simulated dataset from the BKV genome, with a set of reads and a seed contig. To extend, run the following command:
./dist/extender_wrapper --coverage 50 examples/BKV_seed_1000_867.fa examples/BKV_250_50_0.01_0_.fastq
The --complex-threshold -1 option ignore prinseq quality checking for faster execution.
./dist/extender_wrapper --complex-threshold -1 --coverage 50 examples/BKV_seed_1000_867.fa examples/BKV_250_50_0.01_0_.fastq
The output contig(s) for each input will be found in the output/contigs.fasta, sorted by length and in the order they appeared in the input. For example, the output contigs for an input sequence >reference would be named >reference_1 reference, >reference_2 reference, etc. To verify the accuracy of the extended contig, the reference genome is provided in BKV.fasta.
Paired-end example:
./dist/extender_wrapper examples/vir_ref.fa --m1 examples/reads_1.fq --m2 examples/reads_2.fq --enable-pair-constraint
Paired alignments with incorrect orientation or fragment lengths longer than 500bp are excluded.
use higher value for --extend-tolerance [2.5, 3.5] (default 2.5) to force extension, but risks higher chances of false extension.
./dist/extender_wrapper --extend-tolerance 3.0 examples/BKV_seed_1000_867.fa examples/BKV_250_50_0.01_0_.fastq
For best results, preprocess fastq by removing adaptors before attemping extension Filtering based on complexity is an available option (uses DUST method from PRINSEQ)
usage: extender_wrapper [-h] [--m1 [M1]] [--m2 [M2]] [--out [OUT]]
[--enable-pair-constraint]
[--min-overlap-length [MIN_OVERLAP_LENGTH]]
[--extend-tolerance [EXTEND_TOLERANCE]]
[--coverage [COVERAGE]]
[--min-branch-score [MIN_BRANCH_SCORE]]
[--branch-limit [BRANCH_LIMIT]]
[--stop-length [STOP_LENGTH]] [--threads [THREADS]]
[--complex-threshold [COMPLEX_THRESHOLD]]
[--maxins [MAXINS]]
reference [reads]
positional arguments:
reference fasta file with contigs to extend
reads fastq file with unpaired reads to extend with; use the
--m1 and --m2 options instead for paired reads
(default: None)
optional arguments:
-h, --help show this help message and exit
--m1 [M1] fastq file with #1 mates of paired data (default:
None)
--m2 [M2] fastq file with #2 mates of paried data (default:
None)
--out [OUT] output directory (default: output)
--enable-pair-constraint
Require paired-end alignments to satisfy orientation
and insert size constraints according to Bowtie2.
Otherwise, paired data will be treated as two sets of
unpaired reads. (default: False)
--min-overlap-length [MIN_OVERLAP_LENGTH]
minimum length of overlap between candidate read and
the existing contig (default: 15)
--extend-tolerance [EXTEND_TOLERANCE]
This parameter, along with read length and coverage,
is used to determine the required score to extend.
Lower numbers require better quality alignments to
extend (default: 2.5)
--coverage [COVERAGE]
estimate of sequencing coverage (default: 10)
--min-branch-score [MIN_BRANCH_SCORE]
A new branch for an alternative contig will be created
if it meets this threshold. Then, both the main and
alternative branches will be processed recursively.
(default: 5)
--branch-limit [BRANCH_LIMIT]
This limits the number of alternative contigs that are
considered and output. Once the search tree has this
many leaf nodes, no new branches will be created
(default: 1)
--stop-length [STOP_LENGTH]
Terminate extension if any substring of this size is
repeated within the contig. This prevents circular
genomes from infinite extension. (default: 250)
--threads [THREADS] Number of threads to use in computing alignments
(default: 8)
--complex-threshold [COMPLEX_THRESHOLD]
[0-100] This parameter is passed to PRINSEQ's DUST
complexity filter, is run on the reads. Higher values
indicate less complexity. -1 to disable (default: 15)
--maxins [MAXINS] When using paired-end constraints, this is the maximum
fragment length, passed directly to Bowtie2. This
length includes both reads ond the gap between them.
Refer to the Bowtie2 manual for more information
(default: 500)
Install Docker and build the image with
sudo docker build .