We are happy and proud to announce that MatchMiner – the result of our long collaboration with Dana Farber Cancer Institute (DFCI) – is soon going open-source and will be available online. MatchMiner is a computational platform for algorithmically matching patient-specific genomic profiles to precision medicine clinical trials.
Background
In 2015, DFCI initiated a project to develop a platform for the automatic matching of patient-specific genomic alterations to clinical trials. As a large cancer center, DFCI registers hundreds of new cancer patients every year, and each new patient undergoes sequencing with an extensive cancer gene panel. Being a large institution with a large intake, DFCI faced two major issues: oncologists had to manually track dozens or even hundreds of active clinical trials, only a few of which may be relevant to any one patient at a time; and clinical trial investigators had a challenge to manually identify patients eligible for their trials.
Clinical trial investigators
The field of clinical trials has been rapidly developing along with the fast development of precision oncology. When genomic profiles and the information about individual mutations are easily accessible, there is an opportunity to develop a complex multi-arm clinical trial which could have a higher chance of success. However, clinical trials with a more complex structure are often difficult to describe in a way that would be both user-friendly and machine-readable. While this can be difficult for a trial run within one institution, it becomes even more challenging in case of precision medicine basket trials, such as the NCI-Match trial, which are aiming to enroll patients across multiple histologies. The job of a clinical trial investigator to find patients eligible for a clinical trial can become daunting when dealing with genomic profiles of hundreds of patients.
Oncologists
The increasing volumes of available genomics data improve patient outcomes, however, they also put extra administrative burden on oncologists. Individual oncologists have to track numerous active clinical trials, only a few of which may be relevant to any one patient at a time.
MatchMiner as a solution
As an oncologist, when you sit down with your patient, you can see the patient’s profile page with his/her genomic information and a list of clinical trials the patient is eligible for. Patient-specific information includes somatic genomic events, such as:
- mutations
- copy number alterations
- structural variants
Basic demographic and specific clinical data extracted from the Electronic Medical Record (EMR) are also presented:
- cancer type
- age
- gender
In case your patient is eligible for a certain clinical trial and could benefit from participating in it, you can directly contact the clinical trial investigator responsible for this trial.
Technical base of MatchMiner
The software architecture of MatchMiner is divided into two main components to increase developmental flexibility. The MatchEngine is written in Python using the Eve framework to expose a RESTful API. All data is stored and indexed in several MongoDB collections. The frontend is written in AngularJS 1.5 using the Material Design components and philosophy. ElasticSearch is also used to facilitate searching of the data, enabling users to create customized aggregate search queries. MatchMiner supports all major browsers.
What is unique about MatchMiner
MatchMiner is the first fully open-source trial matching platform, comprehensively built and tested within a major NCI-Designated Cancer Center. It is also the first platform to use a standards based, open markup language, Clinical Trial Markup Language (CTML) for structuring clinical trial inclusion/exclusion criteria and trial arms information.
Clinical Trial Markup Language was introduced by DFCI to structurally define the eligibility criteria for a clinical trial. As any markup language, it uses a structured text model and allows to create queries based on specific parameters / filters instead of parsing raw text (and makes it possible to eliminate the step of natural language processing). Using markup language to describe clinical trials criteria – genomic alterations, clinical and demographic information – makes it easy to filter patient information and select patients with genetic / clinical / demographic profiles fitting the clinical trial criteria.
As the platform evolves to meet the needs of cancer-specific genomic sequencing, we anticipate that the platform will be widely adopted and jointly developed with additional national and international collaborators. We are proud to share that MatchMiner has currently won the Harvard Business School Kraft Precision Trials Challenge.
The Hyve, in close collaboration with DFCI, developed the front-end of MatchMiner. If you want to know more about MatchMiner or have any questions, please don't hesitate to contact us!
