NEW YORK – CRISPR-based diagnostic technology developer Sherlock Biosciences will soon debut an assay design service that allows users to access the firm's novel artificial intelligence algorithms and create their own tests. The service will be free and is estimated to reduce assay development time from one year down to three months.
In the past few years, Sherlock has commercialized a COVID-19 test with support from the Bill and Melinda Gates Foundation, acquired instrument-free point-of-care test maker Sense Biodetection, and licensed ambient nucleic acid amplification tech, which it intends to incorporate into CRISPR-based diagnostics.
However, offering customers a platform to develop their own CRISPR-based diagnostics has also been a previously undisclosed part of Sherlock's overall business strategies for the past three years, according to CEO Bryan Dechairo.
Creating diagnostic assays typically requires iterative cycles of designing, building, and testing them.
The idea for using AI-based assay design originally came from the development work of Sherlock Cofounder Jim Collins, Dechairo said, and Sherlock gradually brought on experts from Collins' lab and built up a software development team.
While the firm has been using the "back end" AI method for internal assay design, the "front end" user-facing operating system is new, and recently completed.
Sherlock currently has undisclosed beta users of the tool, Dechairo said, including nongovernmental organizations and others, that are developing assays for illnesses like Lassa fever and tuberculosis. Along these lines, work by Sherlock Cofounder Deborah Hung, in collaboration with other academic groups, has recently led to the development of a novel tiled assay to detect cell-free DNA of tuberculosis, as recently described in Nature Communications.
The firm's CRISPR-based assay design experience helped it generate a wealth of knowledge and data that is in turn utilized in the AI-based method, Dechairo said.
The approach involves using machine learning to evaluate the entirety of possible targets for detection and to design thousands of possible assays that are then assessed in silico.
Dechairo said the approach improves the success rate of high-throughput assay screening and allows users to more quickly generate tests with high sensitivity and specificity.
And "it saves a ton of money because you're not ordering all these reagents and then trying to run things only to find out that they all fail," he said, adding, "You don't have to know anything about CRISPR or chemistries to have all of this done for you automatically."
The interface will allow users to select from not only different hosts or pathogens but also from different CRISPR and amplification chemistries with varying temperature requirements and multiplexing capabilities.
The AI development interface has other advantages, as well, Dechairo said, for example by allowing the algorithm to develop assays for sequence spaces for which there aren't existing assays, rather than just focusing on target regions that are already known.
"What you're getting is the best of all possible assays, not the best of what the researcher thought of today or where the researcher wanted to look today," he said.
Sherlock's team of software developers put the algorithms into an operating system "that enables anybody else to just log in, pick which species they want a diagnostic for, what they want to exclude and include, and then just push the button and design those assays," he said.
The design service will be free of charge, and users can order reagents from Sherlock or elsewhere to evaluate the assays in their own labs.
"Once they validate the assay using chemistry in their own labs and it is working very well, if they want to put that on to their own platform, their own devices, or the systems they already have in place, we are happy to discuss giving them a license to CRISPR since we have the worldwide exclusive rights to our chemistries," Dechairo said.
If they desire, however, users can also work with Sherlock to port the assays onto the company's different testing systems, and Sherlock can supply these for customers to commercialize, Dechairo said.
"We can't boil the ocean and do everything ourselves," he said, so this is "a way for us to really scale."
Sherlock expects to officially launch the AI operating system in October.
In addition to bringing CRISPR-based diagnostics to the global health market, Sherlock is also currently focused on developing home-use tests for sexually transmitted infections.
As part of this mission, Sherlock hired Maurice Exner earlier this year to be the firm's first CSO. With a pedigree from Thermo Fisher Scientific, Quest Diagnostics, Abbott, and Hologic, Exner said in an interview that he is ready to address the challenges of bringing molecular diagnostics to decentralized settings.
The key, he said in an interview, is to develop low-cost, easy-to-use tests that are also highly accurate.
"Frankly, I haven't seen anybody who really ticks all of those boxes in a really effective way," he said, but added that he believes Sherlock's growing toolbox of chemistries and consumables will help the firm get there. "I really do think that over the counter is a place where we can actually play and be successful and be actually ticking all those boxes."
In terms of cost, Dechairo said that the firm's research indicates $40 is the upper retail price limit for home-use over-the-counter testing in high-income countries, which in turn constrains the cost of production. That said, "our platform is low cost, disposable, and is easily under $10 all inclusive," he added.
Sherlock Cofounder Collins also said in an interview that the chemistries Sherlock uses aid in the accuracy of the testing.
While the company has been using Cas13 enzymes in its technology, in February Sherlock was granted the US patent for diagnostic use of the Cas12 enzyme, as well. Sherlock obtained exclusive US rights to a patent from Shanghai-based Tolo Biotech as well as additional intellectual property for Cas12 and Cas13 from the Broad Institute.
Collins said that Cas13 is primarily used to target RNA and Cas12 is primarily used to target DNA. They are both particularly attractive for diagnostics for three reasons, he said.
"They are straightforwardly programmable, meaning that you can design their guide RNA to be very specific to a sequence you'd like to detect," he said, and they also have specificity down to the single-nucleotide level.
The enzymes also exhibit collateral cleavage, so they can be linked to fluorescent chemistries that are unquenched when a linker sequence is degraded. This quirk can be exploited to amplify the signal, and the two enzymes can also be used together to enable multiplexing.
For the home-use STI space, Dechairo said that Sherlock expects to launch its over-the-counter chlamydia and gonorrhea test in high-income markets but is also working with global partners and through its own nonprofit, the 221b Foundation, to bring these tests to low- and middle-income markets.
"We will be the first people bringing decentralized CRISPR-based diagnostics to the globe," Dechairo said.