Diagnostics & genomics · DNA sequencing
XGenomes
A Cambridge, Massachusetts company reading DNA in five letter words instead of one base at a time. Since 2017 it has won six federal Phase I grants worth $2,036,113, and no Phase II.
A genome read in words
“I like to think of the genome as a book. The genome has chapters and the chapters could be the chromosomes,” Kalim Mir told TechCrunch in March 2019, days before his company went in front of the Y Combinator Winter 2019 batch. “Current technologies read it letter by letter. [But] we’re recognizing words.”
The words are short. XGenomes stretches unamplified DNA on a surface, opens the double helix, and washes a library of five letter fluorescent probes across it. Each probe finds its complement, lights up, lets go, and finds it again. A super-resolution microscope logs where each flash happened, to within a few nanometers, and how long it lasted. Machine learning turns that blinking into sequence.
The company calls this Sequencing by Emergence. The name is literal: no probe reads a base. The sequence falls out of the collective behavior of the whole repertoire. The Engine, whose Cambridge building the company works from, puts it in one line: “Existing technologies read the information encoded in DNA as ‘letters’, XGenomes read ‘words’.”
What the method is aimed at
Mir’s complaint in 2019 was throughput, not price alone. A genome then took a couple of days and cost around $1,000, he told TechCrunch, down from the $1 billion the Human Genome Project spent. “If we’re talking about population scale sequencing with millions of people we just don’t have the throughput,” he said. He wanted one to two hours and $100.
The grant applications name a sharper target: methylation. Standard sequencing cannot read a methylated base directly, so laboratories convert the sample first and lose accuracy and input DNA doing it. XGenomes argues that a probe binding a methylated site changes its kinetic rhythm, and that the change repeats across every probe touching the site. One measurement would then return the letter and its modification together, which matters most in liquid biopsy, where the cancer signal is a tiny minority of the DNA in the tube.
The founder
Mir was 52 when TechCrunch profiled him. He grew up in Yorkshire and did not start out in a laboratory. “When I was in school there, I was not into science or tech. I was interested in literature,” he said. Aldous Huxley turned him toward genetics. He studied molecular biology at Queen Mary College, took a doctorate in the mid 1990s with Edwin Southern, whose surface chemistry underpins modern microarrays, and later worked as a visiting scientist in Harvard Medical School’s Department of Genetics alongside George Church.
The publication record supports the depth. Mir is last author on a 2018 paper in the Proceedings of the National Academy of Sciences that mapped and sequenced the genomes of single cells inside a nanofluidic chip. That is the lineage of the imaging idea, not a test of the XGenomes chemistry.
Six Phase I awards, no Phase II
The federal record is the clearest timeline the company has. SBIR.gov lists XGenomes Corp with six Phase I awards, zero Phase II awards, $2,036,113 total, a first award year of 2017 and a conversion rate of N/A.
The titles show the method changing underneath. The 2017 NIGMS award was about authenticating cell lines. The 2018 NHGRI award, “Super-Resolution Sequencing on a DNA Lattice,” was sequencing by synthesis on a packed nanoarray. Then the record goes quiet for six years. It resumes in September 2024 with the probe hybridization approach, and in 2025 three more Phase I grants land inside twelve months: methylation sequencing, a circulating tumor DNA assay and a deep learning localization project.
The 2024 grant sets its own test. In its third aim the company will “construct the synthetic test targets and demonstrate Q50 sequencing of targeted variants of a clinically relevant loci.” That award ran to May 2026. Whether the demonstration succeeded is not in the public record.
What is proven, and what is still claimed
| Evidence | What the record shows | Source type |
|---|---|---|
| Federal grants | Six NIH SBIR Phase I awards, $2,036,113: NIGMS 2017 ($225,000), NHGRI 2018 ($280,000), NHGRI 2024 ($384,418), NHGRI 2025 ($397,599), NCI 2025 ($399,816), NIGMS 2025 ($349,280). Principal investigator on all six: Kalim Ullah Mir. | Public record |
| Phase II conversion | SBIR.gov lists $0 in Phase II dollars and a conversion rate of N/A. Three of the six Phase I awards were made in 2025 and run into 2027, so that window is still open. | Public record |
| Published method | One preprint, “Sequencing by Emergence: Modeling and Estimation,” arXiv 2103.10477, by Nicholas Boyd, Samuel Woodhouse and Kalim Mir, posted March 18, 2021. It sets out a probabilistic model and an estimation algorithm. | Public record |
| What that preprint tested | Its own closing line: “We demonstrate the effectiveness of our algorithm on a variety of simulated datasets.” No experimental sequencing data appears in it. | Public record |
| Peer-reviewed platform data | None found in PubMed as of September 24, 2026. The two hits linked to Mir are 2017 and 2018 nanofluidics and optical mapping papers with academic collaborators. | Not found |
| Patents | Filings assigned to XGenomes Corp cover sequencing through “an emergent property of the binding interactions of a repertoire of molecular probes,” random epigenomic sampling and epigenetic modification detection. CB Insights counts ten filings and one grant issued August 30, 2022. | Public record |
| Claims, 2019 | Y Combinator’s batch write-up: “1000x better accuracy and 10x lower cost than existing technology.” Mir told TechCrunch he could reach a genome in one to two hours for $100. The YC page targets consumable cost below $10. | Company-stated |
| Claims, 2024 to 2026 | Q50 on targeted variants in 2024. Accuracy “over 100x greater than current DNA sequencing platforms” in the 2025 NCI grant. The 2026 site advertises “OneBox for prep and >Q60 sequencing.” No price, ship date or customer. | Company-stated |
| Outside assessment | Keith Robison, surveying sequencing startups in February 2020, filed XGenomes under a heading he called “We Exist and That’s All You Need To Know.” No sequencing trade press coverage was found. | Independent |
Read plainly: the chemistry is described in patents and grant abstracts, the inference math is public, and a federal study section has funded the approach four times since 2024. What is missing is the thing that decides whether a sequencing technology is real, which is data from a real sample, benchmarked against an existing platform, that other people can examine. The one public paper tests the algorithm against simulations, and the accuracy claims have climbed from 1000x in 2019 to Q50 in 2024 to above Q60 in 2026 with no published measurement moving alongside them.
What to watch
- A Phase II award. Six Phase I grants and no Phase II is the most informative number in the file, and the 2025 awards make 2027 the year it resolves.
- The result of the 2024 NHGRI aim, Q50 sequencing of targeted variants, on an award that ended in May 2026.
- Any experimental data, in a preprint, poster or paper, showing reads from a real sample rather than a simulation.
- A ship date, price or named customer for OneBox, which appears on the website and nowhere else.
- A Form D or a disclosed financing. None is on file at the SEC.
In their words
“I like to think of the genome as a book. The genome has chapters and the chapters could be the chromosomes.”
Kalim Mir, founder and CEO, TechCrunch, 2019 · Independent
“If we’re talking about population scale sequencing with millions of people we just don’t have the throughput.”
Kalim Mir, TechCrunch, 2019 · Independent
“We developed a way where we’re sequencing directly on the DNA where we’re not manipulating it except for opening up the double helix.”
Kalim Mir, TechCrunch, 2019 · Independent
“XGenomes is developing technology to read DNA sequence information, with 1000x better accuracy and 10x lower cost than existing technology.”
Y Combinator batch announcement, 2019 · Accelerator
“SEQE promises to deliver accurate, ultra-long, haplotype-phased reads at the whole genome-scale for very low cost within 10 minutes.”
Boyd, Woodhouse and Mir, arXiv 2103.10477, 2021 · Company preprint
“We demonstrate the effectiveness of our algorithm on a variety of simulated datasets.”
Same preprint, closing line of the abstract · Company preprint
“If you want something barely above stealth, check out XGenomes website”
Keith Robison, Omics! Omics!, 2020 · Independent
Related companies
Sources
- Public recordSBIR.gov company portfolio, XGenomes Corp
- Public recordSBIR award 219477, 1R43HG013914-01A1
- Public recordNIH RePORTER, six SBIR awards to XGenomes
- Public recordSequencing by Emergence: Modeling and Estimation, arXiv 2103.10477
- Public recordSingle-molecule DNA-mapping and whole-genome sequencing of individual cells
- Public recordPatents assigned to XGenomes Corp
- IndependentXGenomes is bringing DNA sequencing to the masses
- IndependentA Lazy Look at The Field of Sequencing Startups
- IndependentXGenomes profile, patents and funding
- IndependentXGenomes profile, deal history and patents
- CompanyMeet 5 Companies from the YC W19 Batch Part 6
- CompanyXGenomes company page and description
- CompanyXGenomes website, home and careers pages
- PartnerXGenomes resident company page
Profile researched and written by Healthcare Discovery. Last updated September 29, 2026.
