“Mastermind” Scientists Help Diagnose Rare Diseases
Blog Mar 15, 2019
Rare diseases are problematic for clinicians and scientists as it is extremely challenging to study their etiology. It’s therefore a sad reality that many patients suffering from rare conditions often do not receive a diagnosis, making their care management a difficult process.
The Rare Genomics Institute, a nonprofit organization providing support to undiagnosed rare disease patients, announced on Rare Disease Day 2019 that they had used the Mastermind Genomic Search Engine, a clinical decision support tool by Genomenon, to diagnose a rare disease patient. We spoke to Mark Kiel, MD, PhD, Founder and Chief Science Officer at Genomenon and Romina Ortiz, COO at the Rare Genomics Institute, to learn more about the application of this technology in rare diseases.
Ruairi MacKenzie (RM): How can receiving a faster diagnosis help patients with rare diseases
Romina Ortiz (RO): Unfortunately, the average length of time from when symptoms begin to appear and receiving an accurate diagnosis is nearly 5 years. During that time, patients see an average of 7 different physician specialties. Often when a condition is rare, a patient must travel to a number of different locations across the state, country, and even world, to find the few experts available.
At the same time, rare conditions are less likely to be picked up by typical diagnostic tests. The delay to diagnosis, or what we call the diagnostic odyssey, has been shown to result in inappropriate testing, treatments, and missed opportunities, resulting in increased morbidity (severity of the illness) and mortality.
At Rare Genomics, we have seen first-hand patients dying waiting for a diagnosis. And those that do live are desperate, using the little resources they have to achieve it. A faster diagnosis means giving the families a name to what is causing such harm, it gives a legitimate genetic answer that they can use for a faster route to appropriate covered testing, treatments, and hopefully a cure. It also opens a community for them with other patients with the same diagnosis. This community can share resources, advice, and friendship to families who have felt alone and isolated for so long.
RM: How does Mastermind determine which genetic variants may be relevant to a specific patient?
Mark Kiel (MK): When a patient is diagnosed with a rare disease, their DNA can be sequenced to find the genetic variants that may be causing the disease. The patient’s DNA profile is filtered to find rare variants that aren’t widely found in the population, and each of these variants is researched by the geneticist to see if there is any scientific evidence that ties the variant to a specific disease.
Mastermind accelerates the variant research by putting all the scientific evidence at the geneticist’s fingertips. By searching Mastermind, geneticists get a comprehensive view of all the published research tied to a patient’s variant, the disease context, and deep insight into every article to quickly find the relevant information they need to make a diagnosis.
Molly Campbell (MC): In what ways are the capabilities of Mastermind greater than other genomic analysis tools currently available?
MK: Mastermind uses a high-speed automated process, Genomenon’s proprietary Genomic Language Processing (GLP), and Artificial Intelligence to index the entire corpus of genomic scientific literature. With this process, Genomenon has indexed over 100 times more content than the leading manually curated database and delivers over twenty times the amount of genomic content found by that competitor.
Mastermind has several distinct advantages over manually curated variant databases:
1. Comprehensiveness: Mastermind is the most comprehensive database for genomic literature in the world covering over 6.2M genomic scientific research articles and 4.1M genomic variants. Mastermind is updated weekly to assure that the latest research can be found by the search engine.
2. Speed: The application of technologies like Artificial Intelligence, Machine Learning, and Genomic Language Processing, combined with sophisticated filtering and sorting algorithms, organizes the Mastermind data in such a way that the time it takes to find the research you need is significantly shortened. User feedback shows they can research a patient’s variant 10 times faster than their prior methods.
3. Scalability: With over 500,000 new genomic publications published last year, the growing volume of research is just too much for a manual curation tool to manage. Mastermind’s infrastructure keeps up with the world’s research so that it will always be the best source of evidence-based interpretation.
MC: Considering that some genetic studies in the literature may be of higher quality than others, how does Mastermind determine the reliability of a study from which it is extracting data?
MK: Mastermind seeks genomic data from all known scientific publishers and presents it to the researcher without bias to provide the most comprehensive view of the literature landscape. Mastermind’s first mission is sensitivity - that is to say, comprehensiveness in search results. If there is a paper that contains a patient’s variant, the goal is to find it and display it in the search results.
Mastermind then goes further to ensure the best results rise to the top: the search results are prioritized and displayed by clinical relevance based on the quality ranking of the publication, citations, and relevance to the search terms.
MC: Your press release mentions “technology that wasn’t available just a few years ago”. What newly available technology enables Genomenon to conduct its searches?
MK: Mastermind uses Artificial Intelligence and Machine Learning to read the titles and abstracts of every scientific medical paper published, over 30 million of them! The full text of papers with genomic information is then indexed to develop the most comprehensive view of the genomic landscape. To date, Mastermind has indexed the text of over 6.2 million genetic publications and found over 4.1 million variants in those articles.
The genomic data found in the publications is processed through Genomenon’s patented Genomic Language Processing (GLP). More sophisticated than Natural Language Processing, GLP identifies every way that an author can describe a gene or variant and filters out erroneous information that can be mistaken for genomic data.
The data is then presented in a multi-faceted web interface, utilizing sophisticated algorithms designed to show the most relevant results first. Advanced filtering allows the user to search broadly for maximal sensitivity or narrowly for optimal specificity.
RM: How can journals make it easier for clinicians and tools like Mastermind to access data that can help patients?
MK: There is a great debate between the scientific and publication communities on how to make research more accessible to better help patients with life-saving information. Mastermind bridges those communities by delivering the most comprehensive landscape of the published research associated with a patient’s genomic data. Once the key papers have been identified, Mastermind takes the researcher to the journal website to view or purchase the articles they need to make the final diagnosis.
Mark Kiel and Romina Ortiz were speaking to Molly Campbell and Ruairi MacKenzie, Science Writers for Technology Networks.