Pacific Biosciences Of California, Inc. · NASDAQ:PACB
Pacific Biosciences Of California — Technical Presentation (February 2024)
16 slides from Pacific Biosciences Of California's technical presentation, Feb 21, 2024
Slide 1: Bioinformatics resources to analyze PacBio HiFi human genome
Bioinformatics resources to analyze PacBio HiFi human genome February 21–-22, 2024 | Bioinformatics webinar Billy Rowell | Senior Staff Bioinformatics Scientist 1
Slide 2: Accessing the PacBio WGS Variant Pipeline
Accessing the PacBio WGS Variant Pipeline Available on multiple cloud vendors (DNAnexus, Form Bio, Terra, DNAstack) https://github.com/PacificBiosciences/HiFi-human-WGS-WDL 2
Slide 3: Setting up and running the PacBio WGS Variant Pipeline on-premises
Setting up and running the PacBio WGS Variant Pipeline on-premises Set up the environment: Set up a run: 1) Install WDL execution engine. 6) Prepare data (or optionally, download 2) Install container runtime. example data). 3) Configure your WDL execution 7) Prepare inputs definitions file. engine t
Slide 4: Choose a WDL execution engine
Running on-premises: Choose a WDL execution engine Cromwell miniwdl • Flexible and highly configurable • Simple base configuration • Call caching and workflow recovery • Written in python with a plugin system with database backend • Call caching with file-based backend • Two modes: • Single workflow
Slide 5: Running on-premises: Choose a container runtime
Running on-premises: Choose a container runtime • For improved reproducibility and ease of deployment, all tools are containerized and hosted on Quay. • Cromwell and miniwdl download, manage, and execute tools in these containers with runtimes such as Docker and Apptainer. • While Docker is the most
Slide 6: Running on-premises: Configuring for use with HPC job schedulers
Running on-premises: Configuring for use with HPC job schedulers Job Scheduler Cromwell miniwdl local ✔ Supported by configuration ✔ Supported by configuration Slurm ✔ Supported by configuration ✔ Supported by plugin SGE/Grid ✔ Supported by configuration ✖ Unsupported LSF ✔ Supported by configuratio
Slide 7: Running on-premises: Downloading workflow
Running on-premises: Downloading workflow • If you plan to run analysis on HPC, clone the GitHub repository. • For reproducible analysis, clone only tagged releases (e.g., v1.1.0). • Note: This repository uses submodules, so make sure to use --recursive and to update submodules when pulling updated
Slide 8: Running on-premises: Downloading reference data
Running on-premises: Downloading reference data • The workflow is compatible with the GRCh38 reference. • Reference data (sequence, index, and annotations) have been uploaded to AWS, Azure, and GCP, with pre-populated templates available on GitHub. • For on-premises use, reference data can be downlo
Slide 9: Running on-premises: Downloading test data
Running on-premises: Downloading test data • For each sample, the workflow input is: • data: one or more unaligned HiFi BAMs (~30-fold depth) • metadata: sample identifiers, sample sex , phenotypes, relationships to other samples • Sample Revio data for the GIAB Ashkenazi Trio can be obtained from:
Slide 10: Running on-premises: Set up inputs for your first run
Running on-premises: Set up inputs for your first run • Grab a copy of the inputs.hpc.json template HiFi-human-WGS-WDL/backends/hpc/inputs.hpc.json from the repo you downloaded • In your copy, update paths for all reference data, replacing “<local_path_prefix>” with full absolute path • Delete rows
Slide 11: Running on-premises: Set up inputs for your first run
Running on-premises: Set up inputs for your first run Template Example values for HG002 singleton inputs Detailed descriptions of cohort and sample inputs Phenotypes are Human Phenotype Ontology terms as provided by https://hpo.jax.org 11
Slide 12: Running on-premises: Submit the job
Running on-premises: Submit the job • The exact syntax will vary, but you will typically provide paths for: • The configuration to be used by the workflow execution engine. • The “entry-point” to this workflow: ./HiFi-human-WGS-WDL/workflows/main.wdl • The inputs.json file created for this singleton
Slide 13: Running on-premises: Expected runtimes and output filetypes
Running on-premises: Expected runtimes and output filetypes • In the absence of resource bottlenecks, analysis sample outputs bam_stats cohort outputs cohort_small_variant_vcf runtime is approximately 17 hours (single sample) read_length_summary cohort_sv_vcf read_quality_summary to 22 hours (multi-
Slide 14: Expected variant types and counts
Expected variant types and counts Tool Per-sample mean SNVs and small indels (<50 bp) DeepVariant 4.8M SNVs, 1.1M small indels Structural variants (≥50 bp) PBSV 25k CNVs (>100 kbp) HiFiCNV 49, covering 13 Mbp Phase block NG50 HiPhase (DeepVariant + PBSV) 1.2 Mbp Tool Per-sample mean Tandem repeats T
Slide 15: Updates and improvements expected in the next year
Updates and improvements expected in the next year Maintenance and design improvement Additional analysis types • Workflow changes to improve compatibility • Typing of pharmacogenomics genes (PGx) with platforms • Homoplasmic and heteroplasmic variant calls • Updated versions of: for mitochondrial D
Slide 16: www.pacb.com
www.pacb.com Research use only. Not for use in diagnostic procedures. © 2024 Pacific Biosciences of California, Inc. (“PacBio”). All rights reserved. Information in this document is subject to change without notice. PacBio assumes no responsibility for any errors or omissions in this document. Certa
Pacific Biosciences of California — Technical Presentation (February 2024) on 6ix