We’ve been very busy since our January update, and have a lot to share. As always, if you have questions or see opportunities to collaborate, please let us know; we’re eager to work with others thinking along similar lines.
We’ve continued to collaborate with Marc Johnson’s lab at the University of Missouri to sequence wastewater influent. We’ve also begun to ramp up our in-house process, using the NovaSeq X+ sequencers at Broad Clinical Labs. Over the past three months, we’ve sequenced 270B read pairs from thirteen sewersheds, more than all our pre-2025 sequencing combined. This includes the addition of a sewershed in South Florida, through a collaboration with Helena Solo-Gabriele’s lab at the University of Miami, and a new sewershed from the Midwest with significant input from meat processing facilities, through the Johnson lab. We continue to be enthusiastic about collaborating with additional partners here to expand the scope of the monitoring system.
Our ANTI-DOTE contract work for PHC Global (press release) has expanded to three sewersheds, and now includes wastewater from military bases in addition to marine blackwater.
We’ve written up our key conclusions from our fall 2023 collaboration with CDC’s Traveler-based Genomic Surveillance program and Ginkgo Biosecurity, comparing pooled airplane lavatory waste to municipal treatment plant influent. We plan to release a preprint after our collaborators have reviewed it. While this report considers only the pilot sequencing results from last spring, we’ve now received a full set of prepared aliquots from MIT’s BioMicro Center and have sequenced an initial batch of these at Broad Clinical Labs. We’ll make a decision later this quarter about whether to sequence the full set now, and if so, how deeply.
Last winter, we collaborated with Jason Rothman, Katrine Whiteson, and the Southern California Coastal Water Research Project to sequence wastewater from Los Angeles. While the data has been on SRA since December (PRJNA1198001), we’ve recently submitted a data paper for publication describing the data generation process and giving additional metadata. A preprint is available on SSRN.
Over Q1, we ramped up our swab collection efforts, collecting swabs from 728 people across sixteen collection runs (dashboard). Our goal was to collect enough data to decide to either scale the program up or shut it down, and based on promising results over Q1 we’ve decided to scale it up. We’ll publish a report explaining our reasoning and our overall findings in mid Q2.
To effectively process pools of dry swabs, we’ve needed to develop a nucleic acid extraction protocol, which is public on protocols.io along with our existing wastewater protocols.
Our review on indoor air sampling for detecting viral nucleic acids has now been published in the Journal of Aerosol Science (blog post). This peer-reviewed publication builds upon the preprint we previously announced in May 2024, and includes analysis of public metagenomic sequencing data, a review of sample collection methods, and discussion of implementation strategies. We would be excited to collaborate with others, especially in cases where we could run our pathogen detection algorithms on data collected by partners with air sampling expertise. In particularly promising cases, we could potentially cover sequencing and sample processing costs to make this happen.
We continued to optimize and professionalize our main reference-based analysis pipeline, mgs-workflow. It’s now even cheaper to run, because we’ve converted it to streaming and dropped several redundant stages.
While we had hoped to have reference-based growth detection (RBGD) in production by the end of Q1, we’ve realized that mgs-workflow needs to have better handling of rare and ambiguous sequences before we expect RBGD to perform well. Improving these inputs to RBGD is a high-priority project for us in Q2.
We’ve continued to run our chimera detection system on data as it comes in. The most interesting finding from the last few months has been a cluster of sequencing reads with a junction between cytomegalovirus (CMV) and something synthetic. With help from the Johnson lab we’ve determined these are most likely lab discards from CMV vaccine work.
When we find a suspicious chimeric junction or other unusual feature in the data, we often would like to know more context. In our last update post, we described a new tool for iteratively assembling outward from a seed of interest. This quarter we’ve continued the tool’s development, are now using it internally for investigating chimeric junctions, and have made it open source.
We’re also applying this assembly tool to detect threats without relying on viral reference sequences. We have prototyped a pipeline for flagging short sequences with growing abundance, and are experimenting with assembling outward from these flagged sequences and characterizing the resulting contigs. Mike McLaren presented a poster at the Chemical and Biological Defense Gordon Research Conference, describing these efforts to construct a reference-free detection pipeline.
We published the third blog post in our series evaluating blood as a potential sample type for biosurveillance, analyzing six public datasets to understand the relative abundance of various viruses in metagenomic sequencing data from whole blood and plasma. While we think a comprehensive system would include blood, our current best guess is that it’s approximately third priority, after wastewater and either nasal swabs or air sampling.
We recently hired a Research Scientist, Katherine Stansifer, and a Laboratory Technician, Róisín Floyd-O’Sullivan. Katherine comes to us from Biobot Analytics where she analyzed wastewater sequencing data and developed pipelines to detect SARS-CoV-2 variants. She’s joined our Near Term Detection team to streamline our handling of incoming data, professionalize our pipelines, and generally help us improve our computational detection capabilities. Róisín also comes to us from Biobot Analytics where she processed wastewater samples for pathogen detection and was involved in lab operations management. She’s joined our wet lab team to process and sequence samples for our philanthropic-funded and contract sequencing, and improve our lab organization.
We’ve received a substantial grant from Open Philanthropy. The largest portion, $3.03M, is for scaling up our wastewater sequencing to three NovaSeq X 25B runs weekly. It also includes $350k for improving experimental methods and exploring other detection opportunities, and $50k as a reserve for rapid response needs if we find something seriously concerning in the sequencing data.
The NAO was featured in a recent piece in Transformer, discussing efforts to increase societal resilience in the face of risks from AI.