GoldPolish (aka GoldRush-Edit) is an efficient draft genome assembly polishing tool that uses long reads for polishing. ntEdit polishes the draft assembly and flags additional erroneous regions, then Sealer fills assembly gaps and erroneous sequence regions flagged by ntEdit. The polisher is adapted from the ntedit_sealer_protocol to use long reads instead of short reads.
-
Build
-
Run
- GNU Make
- Python 3
- btllib v1.6.2+
- ntLink v1.3.5+
- minimap2
- snakemake
- intervaltree
The dependencies can be installed through Conda package manager:
conda install -c conda-forge -c bioconda compilers meson ninja boost-cpp btllib ntlink minimap2 snakemake intervaltree
If you use GoldPolish in your research, please cite:
Wong J, Coombe L, Nikolić V, Zhang E, Nip KM, Sidhu P, Warren RL and Birol I (2023). Linear time complexity de novo long read genome assembly with GoldRush. Nature Communications, 14(1), 2906. https://doi.org/10.1038/s41467-023-38716-x
To build GoldPolish and install it at $GOLDPOLISH_PREFIX, run the following commands from within the goldpolish directory:
meson setup build --buildtype release --prefix $GOLDPOLISH_PREFIX
cd build
ninja install
To polish a draft assembly named assembly.fa with long reads named reads.fa and store the results at assembly-polished.fa, run the following:
goldpolish assembly.fa reads.fa assembly-polished.fa
You can run goldpolish --help to see the available options:
usage: goldpolish [-h] [-k K] [-b BSIZE] [-m SHARED_MEM] [-t THREADS] [-v] [-x MX_MAX_READS_PER_10KBP] [-s SUBSAMPLE_MAX_READS_PER_10KBP]
[--ntlink | --minimap2 | --mappings MAPPINGS]
seqs_to_polish polishing_seqs output_seqs
positional arguments:
seqs_to_polish Sequences to polish.
polishing_seqs Sequences to polish with.
output_seqs Filename to write polished sequences to.
optional arguments:
-h, --help show this help message and exit
-k K k-mer sizes to use for polishing. Example: -k32 -k28 (Default: 32, 28, 24, 20)
-b BSIZE, --bsize BSIZE
Batch size. A batch is how many polished sequences are processed per Bloom filter. (Default: 1)
-m SHARED_MEM, --shared-mem SHARED_MEM
Shared memory path to do polishing in. (Default: /dev/shm)
-t THREADS, --threads THREADS
How many threads to use. (Default: 48)
-v, --verbose
-x MX_MAX_READS_PER_10KBP, --mx-max-reads-per-10kbp MX_MAX_READS_PER_10KBP
When subsampling, increase the common minimizer count threshold for ntLink mappings until there's at most this many reads per 10kbp of polished sequence.
(Default: 150)
-s SUBSAMPLE_MAX_READS_PER_10KBP, --subsample-max-reads-per-10kbp SUBSAMPLE_MAX_READS_PER_10KBP
Random subsampling of mapped reads. For ntLink mappings, this is done after common minimizer subsampling. For minimap2 mappings, only this subsampling is done.
By default, 40 if using minimap2 mappings and 100 if using ntLink mappings.
--ntlink Run ntLink to generate read mappings (default).
--minimap2 Run minimap2 to generate read mappings.
--mappings MAPPINGS Use provided pre-generated mappings. Accepted formats are PAF, SAM, and *.verbose_mapping.tsv from ntLink.
--target Run GoldPolish in targeted mode
-l LENGTH, --length LENGTH
GoldPolish-Target flank length (if --target specified) (Default: 64)
--bed BED BED file specifying target coordinates (if --target specified)
--softmask Target coordinates determined from softmasked regions in the input assembly (if --target specified)
--k-ntlink k-mer size used for ntLink mappings (if --ntlink specified) (Default: 88)
--w-ntlink Window size used for ntLink mappings (if --ntlink specified) (Default: 1000)
GoldPolish can be run in a targeted mode, polishing specified regions of the assembly (either by looking for softmasked sequences or by using target coordinates that are specified in a BED file). To run GoldPolish-Target, use the --target flag for the GoldPolish command. Example command that runs GoldPolish-Target with a BED file to specify target regions:
goldpolish --target --bed polishing_coordinates.bed assembly.fa reads.fa assembly-polished.fa