I am not a lucky person in the sense of finding a $20 bill on the sidewalk or winning a raffle. In fact, I have the luck of a person who goes to the keg on its last quarter of a beer or is the last car in a line of speeding cars and gets the ticket. So it came as a shock to me when I returned from vacation to find that the data from an experiment my team and I had executed before I left had yielded an unexpected find from one of our negative samples.
We are a sequencing company. We design assays to identify the most relevant genetic changes in cancer samples. Typically, researchers or clinicians obtain tumor material from fixed biopsy specimens that suffer from various quality issues not found in freshly isolated and prepared material. We therefore use material compromised in this manner during the course of our product development activities in order to design according to the types of materials used by our customers with our assays. That is to say, we aim to improve the performance of our assays with respect to the worst-quality material capable of producing an informative sequencing library.
This past spring, my team and I conducted an experiment with the goal to see if we could improve the complexity of the libraries produced by Anchored Multiplex PCR (AMP™) chemistry with a few small changes to the library preparation protocol. We had already made proof of concept observations that these changes would improve results, but we wanted to verify that these changes were broadly beneficial across multiple different samples and technical replicates.
In the past we had screened through a bank of archival cancer specimens using our flagship Archer™ FusionPlex™ ALK, RET, ROS1 v2 Panel and identified samples that contained ALK, RET, or ROS1 fusion genes. Additionally, we analyzed the quality of the RNA and DNA in this collection using the Archer PreSeq™ QC Assays and identified those samples that were of sufficient quality to produce informative sequencing libraries. We therefore had a bank of ALK, RET, and ROS1 fusion-negative, adequate-quality samples that we could use for panel development. We chose five samples such as these to verify that our protocol change was in fact beneficial.
The protocol change we were proposing, although small, could significantly disrupt some of our customers' normal operations. We therefore wanted to be sure that the proposed change was worth the effort. We do teamwork well at ArcherDX, and from time to time will run large collaborative experiments between R&D groups to generate a large amount of data in a very short amount of time. For this project, we combined the efforts of five R&D scientists from three different groups to test the impact of this protocol change on library complexity. Each scientist was tasked to make libraries using three different FusionPlex protocols and two different samples. Each sample and each protocol would have libraries made in ten replicates. Therefore, a scientist was responsible for producing 60 libraries using our (at that point, upcoming) FusionPlex Solid Tumor Panel and Universal RNA Reagent Kit v2.
I was part of this effort and completed my portion of the experiment just prior to going on vacation. I was a good husband on vacation and did my best to stay out of the loop while we were away. Although I had high hopes for the experiment, it turned out to be much better than I could have hoped for.
Not only did the protocol changes recapitulate what we had seen with our earlier proof-of-concept work, but the combination of two elements had an additive effect that improved our complexity two to five fold! This increase in library complexity, which is directly related to sensitivity, was most pronounced for the worst quality samples. Not only that, but we also found an exceedingly rare CRTC3-MAML2 fusion in one of our inputs in all 10 replicates! I was flabbergasted.
As the data kept rolling in from our big experiment, it kept getting better and better. Not only did we confirm the benefits of our proposed change across multiple samples and technical replicates, but we also found a CRTC3-MAML2 fusion, which is very odd for a lung cancer sample. CRTC3-MAML2 fusions are recurrent in mucoepidermoid carcinomas of the salivary gland, but this fusion is supposed to be prognostic for a favorable outcome (1). How did this fusion come to be found in a lung sample if its presence was associated with negative nodal metastasis?
Nevertheless, at the time I felt like serendipity had struck my scientific life with this turn of events (finding the CRTC3-MAML2 fusion). Lately however, I think our CEO might call it differently. He might say that we weren't lucky to have stumbled onto that fusion; we were prepared. By using our most comprehensive (at that point) fusion panel to test our protocol changes, we set ourselves up to find the unlikely. As a scientist, I relish these moments of discovering the unlikely. I take pride in knowing that our products can help share some of this joy with other researchers and clinicians.
2477 55th Street, Suite 202
Boulder, CO 80301