Home · Research vertical
Genomics & pathogens · early research

Reading pathogens straight from the sample — and asking if AI generalises there too.

Metagenomic next-generation sequencing (mNGS) sequences everything in a clinical sample at once, so it can name a pathogen without being told what to look for. We're building an analysis pipeline around it — and testing whether the same generalisation gap, label-noise and calibration problems we mapped in imaging reappear when the signal is a genome instead of a pixel.

The mNGS analysis pipeline

From clinical sample to pathogen & resistance call.

An unbiased, assembly-aware pipeline: deplete the human host, classify what's left against reference genomes, and read antimicrobial-resistance gene content — with the same de-identification and reproducibility discipline as our imaging work.

Sample blood · BAL · CSF Extract + host-deplete Sequence NGS reads QC + filter trim · dehost Classify taxa vs refs Pathogen + AMR call → report De-identified & reproducible end-to-end · every stage version-pinned
Hypotheses under investigation

The questions we're setting out to answer.

This vertical is early — so these are stated as falsifiable hypotheses, each with the leave-one-run-out validation that will prove or reject it. That's the point of the ledger: publish the question before the answer.

◷ Open

The generalization gap has a genomic twin.

A pathogen classifier trained on one sequencing platform, depth or centre will under-detect on another — a batch/depth shift analogous to acquisition shift in imaging.

Test: platform-invariant normalization + multi-run training, scored leave-one-run-out.
◷ Open

Assembly-free profiles catch low-abundance pathogens earlier.

K-mer and embedding representations of raw reads flag rare organisms below the depth where de-novo assembly succeeds.

Test: sensitivity vs abundance, assembly-free vs assembly-based, on spiked and clinical samples.
◷ Open

AMR gene content predicts resistance phenotype before culture.

Resistance-gene signal in mNGS reads anticipates the antibiogram, shortening time-to-appropriate-therapy.

Test: genotype→phenotype concordance against paired culture results.
◷ Open

Unbiased sequencing finds what a targeted panel misses.

Host-depletion + untargeted sequencing detects co-infections and off-panel organisms that syndromic panels cannot, without prior hypothesis.

Test: discordance analysis vs targeted panels on the same samples.
Health markers & multi-omics · exploring

Cheap signals, fused well, may beat expensive ones alone.

Routine blood counts, inflammatory markers and molecular assays each carry partial information. The open question is whether fusing them — and fusing them with imaging — produces triage and stratification that no single modality reaches, at a cost that works in real clinics.

◷ Open

Imaging + a compact marker panel triages better than either alone.

Fusing the chest-imaging reader outputs with routine blood and inflammatory markers improves early triage over the best single modality.

Test: multi-modal fusion vs each modality, calibrated and scored leave-one-site-out.
◷ Open

Marker trajectories stratify treatment response early.

Longitudinal biomarker trends separate responders from non-responders sooner than endpoint tests — for example in tuberculosis therapy.

Test: early-trajectory models against confirmed outcome labels.
◷ Open

Widely-available markers can approximate costly assays for screening.

For triage — not diagnosis — inexpensive, ubiquitous markers may stand in for expensive assays with acceptable loss.

Test: screening-operating-point equivalence with explicit miss-rate bounds.
◷ Open

The same calibration discipline transfers across omics.

Per-site calibration and abstention — proven in imaging — make marker and genomic scores equally trustworthy across populations.

Test: cross-population calibration error before vs after, per vertical.

These are research hypotheses under active investigation, not validated capabilities or clinical claims. Under our proven-capability rule, nothing is adopted until it measures well on data it has never seen.

Work in sequencing, microbiology or biomarkers?

These questions need paired data and clinical ground truth — the kind of collaboration that turns a hypothesis into a result.