Armed with a recently awarded five-year $20 million contract from the US Food and Drug Administration to support its work on the PrecisionFDA platform, DNAnexus is working with Deloitte to enhance the platform’s capabilities for sponsor-reviewer interaction, add support for multi-omics, and provide a library of analytical, statistical, and machine-learning applications.
The features will support the FDA’s “evolving needs for data stewardship, security, and quality, and the increasing complexity of bioinformatics in regulatory review,” DNAnexus CEO Richard Daly noted in a statement. Meanwhile, the platform’s developers are preparing to launch a new community challenge next year that will focus on addressing tumor mutational burden across different gene panels in partnership with Friends of Cancer Research, a non-profit cancer research and advocacy group.
DNAnexus has been developing the PrecisionFDA platform since its launch in 2015 with support from an $849,000 research and development contract from the FDA’s Office of Health Informatics. The platform brings test developers, standards-making bodies, pharmaceutical and biotechnology companies, healthcare providers, and academic medical centers into a collaborative forum to assess the quality of testing methods and create standards around genomic data analysis. “It is important to recognize the way they have thoughtfully progressed and built community involvement and engagement with sponsored companies,” Daly said in an interview. “It’s really one of the major projects that they have on what we think of as the digital transformation of the FDA and using cloud resources.”
About three years after launching the platform, the FDA began engaging with the biomedical community through computational challenges that tackled needs in the genomics space. Challenges so far have tested the reproducibility of pipelines for processing next-generation sequencing data, the accuracy of variant calling pipelines, and the performance of NGS technology for detecting rare variants or variants with low allele frequencies. Other challenges have tackled approaches for detecting food-borne pathogens and biothreats, and algorithms for accurately detecting and correcting mislabeled samples and data.
Currently, the PrecisionFDA team is running a challenge focused on using multi-omics data to identify biomarkers and predict brain cancer patient outcomes that will run through February 2020. Next year, the team plans to launch another challenge, designed in partnership with Friends of Cancer Research, that is assessing issues of tumor mutational burden. The goal of this particular challenge is to support the development and evaluation of methods for somatic variant calling and tumor mutational burden estimation. It aims to harmonize panel-based tumor mutational burden measurements to a reference whole-exome sequencing data-based standard. Participants will have access to data from paired tumor-normal cell lines, which they will use to test their assays and analysis pipelines.
To help with these challenges, DNAnexus developed the PrecisionFDA Challenge Construction Toolkit, which the FDA and its partners use to design and deploy community challenges. Specifically, the toolkit enables the FDA to independently develop the necessary assets needed for challenges and then deploy them on the platform. The variety of challenges shows “the pathway that the FDA is looking for,” Omar Serang, DNAnexus’ chief cloud officer, explained. “They want to incorporate multi-omics [including] proteomics, metabolomics, [and] other things, that are going to require regulation and informatics.”
To that end, some of the new funds will support the development of additional applications and omics datasets on PrecisionFDA, according to Serang. Currently, the platform has a lot of reference datasets and tools for genomics but moving forward, “we are going to be seeing now tools for proteomics [and] for metabolomics,” he said. “It will essentially expand the languages that PrecisionFDA speaks to move beyond not just genomics.”
The funds will also support efforts to enable the platform to support multi-center collaborations and to develop functionality for performing interactive reviews, Serang added. Currently, the interactive review process and tools are going through pilot testing. With the new funds, “one of the things that we are going to be investing in is taking that interactive review out of a research phase and making it a very real part of the review process,” he said.
As part of this process, the company is building out tools to help reviewers more easily and quickly assess analysis pipelines. The company is also making improvements to the platform to refine the user experience. “What we are trying to do is provide tools for the reviewers so that they don’t have to be bioinformaticians and programmers but they can actually use these tools, for instance doing a variant comparison,” Serang explained. “A reviewer should be able to take two VCF files and put them into a variant comparator and assess whether the sponsor’s pipeline performs suitably well compared to the industry standard. That’s something a reviewer should be able to do without bioinformatics support.”
PrecisionFDA’s developers also said that the platform has enabled significant advances in regulatory science for next-generation sequencing-based drugs and devices. Serang pointed to the results of the 2016 PrecisionFDA Truth Challenge, which highlighted the accuracy and performance of analytic pipelines developed by companies Sentieon and Verily Life Sciences, among others. “Over the past five years, PrecisionFDA has played an integral role in precision medicine and has quickly emerged as the gold standard resource for germline variant comparison,” Serang said in an email. The platform “provides a secure environment for FDA reviewers to efficiently evaluate sponsor submissions that incorporate germline variant calling informatics.”
Working on the PrecisionFDA platform has helped grow DNAnexus’ business and platform in a number of ways, according to Daly and Serang. For one thing, the company has been through the Federal Risk and Authorization Management Program authorization process, which is designed to ensure effective repeatable cloud services for use by the government. “That required in some ways a fundamental re-architecting of our platform because of the security and compliance characteristics,” Daly said.
The partnership has also provided DNAnexus with “deep insights into the FDA’s objectives for regulating bioinformatics in drugs and devices,” Serang noted, and that is “a very interesting spot for DNAnexus to be in because it gives us a unique view into the world of regulation and discovery.”
This is important because the company has clients working on drug development, target discovery, and biomarker discovery projects. Last year, DNAnexus closed a $68 million Series F venture funding round that it said would support efforts to deploy DNAnexus Apollo, its cloud-based platform, for multi-omics and clinical data science exploration, analysis, and discovery. According to the company, the platform is intended to help pharmaceutical research and development teams quickly test hypotheses and learn more about mechanisms of action, biomarkers, and targets.
The company also has agreements with a number of healthcare networks and academic medical centers. This includes a partnership with Sutter Health network and the University of Pittsburgh Medical Center focused on multiple sclerosis research. That study is expected to analyze de-identified clinical and genomic data from more than 3,000 patients. DNAnexus’ infrastructure also supports the St. Jude Cloud, a public repository of over 5,000 anonymized whole-genome sequences, 5,000 whole-exome sequences, and 1,200 transcriptomes from pediatric cancer patients.