Wednesday, May 15, 2019

Rapid Advances in Germline and Tumor Genomic Information, Services, Companies

In an excellent posting at Dark Daily on May 15, 2019, Andrea Downing Peck reviews expansion of the Cancer Gene Consensus, part of the UK-based COSMIC program (Catalogue of Somatic Mutations in Cancer).
  • See Dark Daily article here.
  • See COSMIC main website here.
  • See website for Cancer Gene Consensus (CGC) here.
  • See an October 2018 article on CGC by Sondka et al. at Nature Reviews Cancer, here.
  • Together, COSMIC and CGC provide a wealth of tabular, searchable, and graphically presented information on oncogenes both germline and somatic.  If you haven't heard of these resources, worth looking at their content and approach.

Larger Context

Public Databases
  • Association for Molecular Pathology manages a website of Validation Resources and Interpretation Guidelines, online here and here.
    • See also Oza et al. at PubMed and, in the right-hand panel, publications PubMed sees as similar.  Find Oza et al. here, go straight into listing of its related articles here.
  • Increasingly, the FDA will be able to use public databases as part of genomic test clearance/approvals: see April 2018 FDA guidance here.
CancerMine: Public Literature Database

A searchable database based on public oncology literatur is CancerMine.  See a new publication by Lever et al. in Nature Methods, May 2019, here.   See the open access database here,  See references early in the Lever et al. paper to CGC  and COSMIC, but also to Network of Cancer Genes, IntOGen, TCGA, ONGene, TSGene, and CIViC.

Increased Use of Functional Data in Clinical Diagnostics

Increasingly, genomic interpretation involves big databases, clinical curation, but also goes back and forth between the interpretation suite and an actual wetlab for molecular studies (such as rapid in vitro CRISPR studies, Moses et al. open access here, and transcriptome sequencing as in Cummings et al., 2017 here, and Kremer et al. 2017 here, Gonorazky et al., 2019, here).

In parallel, look for more attention to intronic mutations or intronic epigenomic changes like methylation, coupled with transcriptomic studies to resolve novel driver or risk mutations - see Dvinge 2015 here, Pique 2019 here.

Pipeline Analysis Coming Into Its Own

In addition to setting up clinical variant databases per se, how we get from the patient sample to the variant report follows a long and complex genomic pipeline.

Few papers have compared approaches along the whole A-to-Z clinical pipeline.   This spring, AstraZeneca in collaboration with genomic software company SolveBio published a first-of-its-kind review of the complete clinical variant pipeline, from raw BAM file to final report, for four different clinical ctDNA tests.
  • The website for SolveBio here.
  • See article by Stetson here
  • Coverage in Genomeweb here.
Note than an EVP at AstraZeneca recently joined the board of directors of Guardant, here.

Machine Learning for Whole Genome Analysis in the NICU

In April 2019, Clark et al. from Rady Children's Hospital and collaborators published a landmark paper usig machine learning to automate phenotyping and interpretation.  See the Science Translational Medicine article here.

This project was picked up for a laudatory blog by NIH director Francis Collins, here.  (See also Genomeweb here, and a trade journal highlighting the incorporation of natural language processing, here.) 

See a 25 minute May 2019 conference presentation online, by program director Dr. Stephen Kingsmore, here.

The network of collaborators on the Rady paper includes Fabric Genomics, Diploid, Alexion, Codified Genomics, Tessella, and Clinithink.  Sequencing via Illumina NovaSeq.

Mix of Public, Institutional, and Commercial Sources

The field of genomics advances by a mix of platform advances, clinical databases, novel chemistry studies (e.g. transcriptomics for variant calls to reduce VUS), and commercial companies.   I've called the field "Digital Genomics" as a nascent industry, in a December 2017 article in Journal of Precision Medicine (here, here).  In that 2017 article, I laid out a possible conceptual map of digital genomic industry players as here:

click to enlarge; demonstration chart from 2017

PierianDx Webinar Online: A Company Applies Digital Genomics to Clinical Sign-outs

A few weeks ago, PierianDx (with whom Illumina established a formal collaboration last winter, here), held a webinar on its 2019 suite of services to support clinical sign-outs.
  • The video, as well as the deck and the transcript, are online at Pierian here.   
  • Worth checking out.
PierianDx integrates a suite of databases, sources, and services into a licensed service.   The webinar is designed to show genomic clinicians how PierianDx can support them, but it's also worth reviewing as a model and case study of how the genomics industry structure is rapidly evolving through a mix of public database, wetlab, platform, and SaaS resources.

The TruSight 500 Illumina panel is under FDA review as a breakthrough device (here).



PubMed Expertise.  When you look up an article on PubMed, the website shows you "related articles" on the right-hand margin.  ubMed lists several related articles, below which, it gives you little text hyperlinks to "see reviews" (that are related) or "see all" (related articles in their own screen).

It's not a direct fit with the other digital genomics topics in this blog, such as SolveBio and Pierian, but see a very nice article from the consultancy DeciBio on digital pathology and spatial digital analysis in pathology for immuno-oncology - May 2019, here.  They also flag about ten relevant abstracts from the upcoming ASCO2019 meeting.