GenomeBoost is a high-speed genome analysis software that produces results identical to those from the GATK Best Practices pipeline, while achieving over 10× faster performance on most CPU-based servers — without requiring additional hardware such as GPUs or FPGAs. GenomeBoost is the core software of GenomeBoost Server™.
GenomeBoost is optimized based on the BWA implementation under the Apache License and integrates key functionalities of GATK, including MarkIlluminaAdapters, MergeBamAlignment, MarkDuplicates, BaseRecalibrator, and ApplyBQSR. The current version of GenomeBoost supports short-read data ranging from 75 bp to 300 bp.
GenomeBoost is proprietary software developed by Genome4me, Inc.
Please refer to the LICENSE file for details on the licensing policy.
The GenomeBoost BWA module is based on the BWA source code under the Apache License.
- Index the reference FASTA file.
$ genomeboost index <reference fasta>
- Execute GenomeBoost to produce an analysis-ready SAM/BAM file.
$ genomeboost preprocess --mark_illumina_adapter --fix_alignment --mark_dup \
--bqsr <reference fasta> <read1.fastq> [<read2.fastq>]
- The
--mark_illumina_adapteroption corresponds to the GATK MarkIlluminaAdapters module. - The
--fix_alignmentoption corresponds to the GATK MergeBamAlignment module. - The
--mark_dupoption corresponds to the GATK MarkDuplicatesSpark module. - The
--bqsroption corresponds to the GATK BaseRecalibrator and ApplyBQSR modules.
The preprocess command automatically configures the number of threads and
memory usage based on the available system resources.
Users can override these defaults using the --threads option to set the number of threads,
and can control memory usage with either the --high-memory or --low-memory options.
The --high-memory option allows GenomeBoost to use more memory for faster performance
(at the risk of running out of memory), while the --low-memory option limits memory usage
to reduce the risk of out-of-memory errors.
The preprocess command also accepts standard BWA-MEM options such as -Y or -K.
For a complete list of available commands and options, run:
$ genomeboost --help
$ genomeboost preprocess --help
For input FASTQ files larger than 30x human WGS data, GenomeBoost may need larger memory than 64 GB. Because GenomeBoost leverages most of the system memory to execute faster, only a single GenomeBoost process should be executed at a time.
Please make sure that there is enough free disk space (at least 4× the size of input FASTQ files) because GenomeBoost stores intermediate data into temporary files.
By examining >430 human WGS datasets, we confirmed that GenomeBoost produces
SAM output, mark-illumina-adapter metrics, mark-duplicate metrics, and
BQSR report files identical to those of the GATK best-practice pipeline.
In our test, we used the WARP (WDL Analysis Research Pipelines) implementation
of the GATK best-practice and the following software versions:
- WARP WholeGenomeGermlineSingleSample v3.3.4
- GATK 4.6
- Picard 3.2.0
- BWA 0.7.18
Here are the detailed steps of the GATK pipeline we tested:
- FastqToSam
- MarkIlluminaAdapters
- SamToFastq
- BWA
- MergeBamAlignment
- MarkDuplicatesSpark
- BaseRecalibrator
- ApplyBQSR
The GATK WARP pipeline executes BaseRecalibrator and ApplyBQSR steps in multiple
processes by dividing the whole genome into tens of intervals to run faster.
However, the output results of the multi-process BaseRecalibrator are
slightly different from the output files of a single-process BaseRecalibrator
because of random numbers generated in the middle of the process.
To remove the random fluctuations in our test, we executed the GATK BaseRecalibrator and
ApplyBQSR in a single process and confirmed that the output files of GenomeBoost
are identical to those of the GATK BaseRecalibrator and ApplyBQSR.
If you have any questions about this software, please contact support@genome4me.com