"A Platform for the Discovery of Novel Microbial Natural Product Chemical Scaffolds and Their Biological Targets"
Heading to Stanford next week for #SDDS2026.
In 7 months, Evoquant has built:
- A 5.2M biosynthetic gene cluster atlas
- An AI platform predicting natural product–target interactions
- A wet-lab pipeline with confirmed bioactivity in 6/6 validation cases (15 in progress)
🧵
We're looking to connect with companies interested in partnering to advance our compounds, out-licensing from our pipeline, and investors exploring the space.
Whether you want to talk business or just talk science, find @chasingmicrobes.bsky.social at the Corporate Poster Session (Apr 27–28).
"A Platform for the Discovery of Novel Microbial Natural Product Chemical Scaffolds and Their Biological Targets"
Heading to Stanford next week for #SDDS2026.
In 7 months, Evoquant has built:
- A 5.2M biosynthetic gene cluster atlas
- An AI platform predicting natural product–target interactions
- A wet-lab pipeline with confirmed bioactivity in 6/6 validation cases (15 in progress)
🧵
3/3
We are actively pursuing some leads from the first batch of strains but are excited about this next batch!
2/3
Now:
Fast/Responsive, interactive view of all 5-million BGCs
- Search/Highlight BGC(s) of interest
- Lasso-tool around clusters to select BGCs
- BGC files fetched from "the cloud"
- BGCs aligned via all vs all DIAMOND
- Displayed in interactive BGC synteny plot
1/3
We now have an atlas of >5 million BGCs, all clustered by similarity.
Our predictive software finds a BGC it says we should pursue.
But exploring/contextualizing these at scale in real-time was still slow and painful.
BGC comparison map but some BGCs are in reverse compared to others
BGC comparison map with all the BGCs aligned
One-button, instant auto-align BGCs
😍
#secmet #NaturalProducts #genetics #microbiology
Note that the clustering was a quick algorithm and hasn't been optimized in the slightest because it's not needed at the moment (inspecting some of these shows that it may be an underestimate of diversity)
Evoquant is planning (registered today) to attend the Stanford Drug Discovery Symposium (SDDS 2026), with @chasingmicrobes.bsky.social presenting a poster.
First time attending and looking forward to it.
Graph depiction the relation of 5 million BGCs
The biosynthetic gene cluster universe (5 million BGCs)
Photo of PCR plate with chemical extracts
First extracts went out for antibiotic testing this week. Still waiting for some additional pipettes from Thermo Fisher (to the point they are sending loaners in the meantime) at which point we can finally start our anticancer and immune modulator assays.
Photo of PCR plate with chemical extracts
First extracts went out for antibiotic testing this week. Still waiting for some additional pipettes from Thermo Fisher (to the point they are sending loaners in the meantime) at which point we can finally start our anticancer and immune modulator assays.
Gave it a chunk of open source code and asked it to clean it up
Given the user says both should be int, and chembl_id is a string, the molecule uid should likely be molregno. But in load_molecules, chembl_id is renamed to uid and used throughout. Let me check the compound section where ChemblMolecule(properties={"uid": row.chembl_id}) is used — that's a string. The user wants int, so these should use molregno instead, or the chembl_id should not be renamed. Actually, re-reading: the user is just saying the types are int — meaning str() wrapping on TAXID uid is wrong, and the chembl_id rename to uid in molecules is wrong (should keep molregno as uid). But wait — in the original code, molecules used chembl_id as uid too. And in compounds, ChemblMolecule(properties={"uid": row.chembl_id}) uses the string chembl_id.
Claude is in a tizzy
Alt text: Line chart showing a two-year price history for a computer component, with prices around $250–$320 through 2024–2025 before sharply rising above $1,100 by early 2026.
Not enough people seem to be concerned about the cost of owning computers right now.
The price of the RAM in Evoquant's workstation computer:
Photo of bacteria colonies growing on agar
Evoquant's first set of bacterial strains are growing. We now have several strains in culture as we begin evaluating our novel strategy for obtaining bioactive and novel natural products.
Photo is of one of the strains growing on a minimal agar media.
Evoquant received SBIR funding from the NIH in 2025 to develop new/better methods for finding anti-cancer and antibacterial small molecules produced by bacteria
With success, we would apply for a Phase 2 grant, which would allow the company to stay alive & commercialize. Unfortunately...
Evoquant is planning (registered today) to attend the Stanford Drug Discovery Symposium (SDDS 2026), with @chasingmicrobes.bsky.social presenting a poster.
First time attending and looking forward to it.
The SBIR program has not been renewed by Congress and therefore THERE ARE NO SBIR applications at NIH or elsewhere. NIH can't even issue "noncompeting continuation awards for existing projects until the SBIR/STTR program is reauthorized."
America is giving up on small business and Science.
Evoquant received SBIR funding from the NIH in 2025 to develop new/better methods for finding anti-cancer and antibacterial small molecules produced by bacteria
With success, we would apply for a Phase 2 grant, which would allow the company to stay alive & commercialize. Unfortunately...
Otherwise, it resolves to classic. Between versions of the AWS CLI, auto resolution criteria may change. To guarantee use of a specific transfer client, set the preferred_transfer_client configuration variable to the appropriate transfer client listed below. classic - Use the builtin, Python-based transfer client that supports all s3 commands, parameters, and most configuration values. crt - Use the AWS Common Runtime (CRT) transfer client when possible. It is a C-based S3 transfer client that can improve transfer throughput. Currently, the CRT transfer client does not support all of the functionality available in the classic transfer client. The list below details what functionality is currently not supported by the crt transfer client option and the corresponding behavior of the AWS CLI if it is configured to prefer the crt transfer client: S3 to S3 copies - Falls back to using the classic transfer client Region redirects - Transfers fail for requests sent to a region that does not match the region of the targeted S3 bucket. max_concurrent_requests, max_queue_size, multipart_threshold, and max_bandwidth configuration values - Ignores these configuration values.
I have been struggling with transfers in AWS between EC2 and S3 being way slower than the advertised EC2 bandwidth
Last week I learned about:
aws configure set s3.preferred_transfer_client crt
and now transfers are super duper fast.
The more you know
Screenshot of evoquantbio.com showing the text: EvoQuant LLC AI & Bioinformatics for Next-Generation Natural Product Drug Discovery Contact Us
🤩 evoquantbio.com 🤩
Public Health Relevance Statement This research addresses challenges in predicting bioactivity in natural product drug discovery by leveraging advanced AI, computational biology, and multi-omics knowledge graphs. This approach has the potential to accelerate the discovery of new, effective drugs for treating a range of conditions, including cancer, infections, and immune-related diseases, by identifying promising leads that could inform the next generation of therapeutics.
🚀 Evoquant has been awarded a $300k Small Business Grant from the National Institutes of Health for our project:
"Leveraging Large-Scale Bioinformatics & AI for Natural Product Drug Discovery and Target Selection"
Hitting the ground running 🏃and excited for the year to come!