CHICAGO – A "fully open-source" application for multiomic variant curation in both research and clinical settings is ready to be translated to clinical practice following the recent publication of a paper describing the platform, according to its developers.
Developed by a coalition led by bioinformatics startup Quantori and the Harvard-affiliated Forome Association, the software, called Annotation and Filtration for Sequencing Analysis (AnFiSA), produces visualizations to help clinicians without an informatics background understand Mendelian genetic variation. Forome, which stands for forum of omics, is an effort to crowdsource discovery of genotypic biomarkers that could be useful for diagnostics and drug development.
The developers and some early users described the technology in a paper last month in the Journal of Biomedical Informatics.
AnFiSA developers said that their software represents "a major step forward in translational medicine" because it incorporates both research and clinical workflows as well as crowdsourcing activities.
AnFiSA is the first major implementation of the Forome Platform, technology that has grown up around the Forome Association, which grew out of the Brigham Genomic Medicine program at Brigham and Women's Hospital in Boston. Quantori is the primary contributor to the open-source community that supports the Forome Platform; other members include Harvard University, IBM, and Oracle.
According to Cambridge, Massachusetts-based Quantori, AnFiSA is meant to bring genetic analysis into routine clinical practice as well as to convince payors that genetic sequencing is medically necessary.
According to the paper, AnFiSA was designed to "enable the seamless transformation of research workflows into novel clinical protocols by building a notebook-like environment for the design and iteration of clinical analysis." The researchers called the platform "simple enough to be used by clinicians while at the same time capable of transforming research workflows into more formal clinical protocols."
The software can support analysis of next-generation sequencing of whole genomes, whole exomes, and gene panels. It features cohort analysis for researchers looking for biomarkers for undiagnosed hereditary diseases.
Quantori CSO Yuriy Gankin said that the AnFiSA framework allows clinicians and other scientists without a bioinformatics background to develop diagnostic guidelines for rare diseases by setting inclusion and exclusion criteria. Those who are trained in programming have a software development environment where they can easily add new features and find and remove bugs.
Michael Bouzinier, a senior research engineer at Harvard who cofounded Forome, said that the debugging functionality provides key data on provenance. "In case some analysis missed a variant, we can find out why this variant was missed and not included in a report for patient," he explained. This leads to more accurate diagnoses in the future.
AnFiSA is meant to fill a perceived hole in the marketplace.
"Despite genomic sequencing rapidly transforming from being a benchside tool to a routine procedure in a hospital, there is a noticeable lack of genomic analysis software that supports both clinical and research workflows as well as crowdsourcing," the paper said. Crowdsourcing is achieved by allowing users to create secondary workspaces within AnFiSA for sharing de-identified data.
Bouzinier said that other tools either only solve "niche problems" such as a process for variant calling or are commercial products that are not truly open source.
"Often, the source is nowhere to be found. It's not on Github," Bouzinier said. Sometimes, the source code is on Github, but the software cannot be installed without paying for support, which may or may not even be offered by the developer.
The authors of the Journal of Biomedical Informatics article said that clinical germline genomics analysis software tends to include proprietary features, often to the detriment of end users.
"Many proprietary commercial curation tools use opaque machine learning and artificial intelligence models to deliver the treating clinician a conclusion while often sacrificing explainability and therefore traceability of the results," they wrote. "Others take explainability quite seriously, nevertheless giving clinician little room for adjudication."
Grant-funded software tools developed for academic research "do not work and are not used outside of the specific hardware and environment used by the original developers," according to the paper.
"Taking all the above in consideration, it would be fair to say that the clinical community working with complex clinical cases such as undiagnosed diseases with unknown genetic etiology does not have an optimal set of supporting tools."
In his previous role as director of informatics for Brigham Genomic Medicine, Bouzinier was tasked with turning internally developed software tools into viable products. "I spent a year actually going from vendor to vendor of open-source tools and asking, 'Can you help us to put our software on your platform?'" he said. "And almost nobody was able to help."
Thus, the Forome Association, Quantori, and their partners decided to develop open-source software designed to fit into a "clinical-grade diagnostic workflow," according to the paper.
As it stands now, AnFiSA can integrate with downstream clinical software like laboratory information management systems and electronic health records by virtue of the Fast Healthcare Interoperability Resources (FHIR) standard.
The current version of AnFiSA does lack structured phenotypic data about each patient, though users can manually input "limited" phenotypic data for pedigree analysis, according to the authors.
Gankin said that Quantori is contributing to AnFiSA on a pro bono basis, but the participation will allow the firm to commercialize specialized services for for-profit entities like pharmaceutical companies. "We will still be able to provide free services to hospitals and not-for-profits, but there are some plans to provide paid services and special versions for the for-profit organizations in the interest of growing," he said.
Bouzinier said that the AnFiSA team identified several basic use cases for AnFiSA in genomic analysis, including traditional genotype-based analysis, cohort exploration for identifying new variants, and newborn screening.
Brigham Genomic Medicine has used AnFiSA to diagnose patients who lived for years with undiagnosed illnesses and to develop a protocol for detecting hearing loss in newborns as part of the Sequencing a Baby for an Optimal Outcome (SEQaBOO) research study; SEQaBOO is listed as an author of the Journal of Biomedical Informatics article.
Researchers at Brigham have also used AnFiSA in the Harvard-led Undiagnosed Diseases Network clinical site and in a cohort analysis of patients with the rare acute skin disorder purpura fulminans.