PerMED
Transcriptome-guided therapy ranking

When the
guidelineruns out

A DNA panel tells you what is broken. It is silent for most patients once the listed options are spent. PerMED reads the whole transcriptome of the tumour and ranks systemic agents by how strongly each one's target pathway is actually running.

Request clinical access See a worked case
PM-DEMO-001
Metastatic colorectal · 2 prior lines
Synthetic
Pathway activation level, 0–100. Bar colour denotes target pathway family and is consistent across cases.
Why the transcriptome

A mutation is a hypothesis. Expression is the evidence.

Most tumours have no actionable mutation

Sequencing DNA alone leaves a large share of patients with no matched option — the panel reports variants of unknown significance, or nothing at all.

Pathways carry the signal

A pathway can be driven hard without a mutation in it. Scoring activation across the transcriptome finds drivers that a variant list cannot see.

Every rank is traceable

Each level decomposes into the genes and interactions that produced it. A tumour board can interrogate the ranking rather than take it on trust.

From block to report

Four stages, one specimen

01

Specimen

An archival FFPE block. A pathologist confirms the diagnosis and marks a region of at least 50% tumour cells for dissection.

02

Sequencing

Paired RNA and DNA from the same dissected region: ribo-depleted RNAseq at ~30M reads, exome or cancer panel at ~90–500× on target.

03

Pathway analysis

Expression is normalised against reference tissue and resolved into activation levels across annotated signalling pathways.

04

Ranked report

Agents are ordered by the activation of the pathways they act on, each with the contributing genes and the literature behind them.

Omics layers

One block, three layers of data

RNA carries the signal the ranking is built on. DNA corroborates it. Both come from the same dissected region of the same specimen, so nothing has to be reconciled across blocks, sites or timepoints.

Layer 01 · RNA

Whole transcriptome

Strand-specific, ribosome-depleted, random hexamer-primed RNAseq, prepared for formalin-fixed material rather than adapted to it. This is the primary input: expression across the coding transcriptome, not a selected gene set.

Library
Ribo-depleted, strand-specific
Depth
~30 million raw reads per sample
FFPE QC
DV200 qualification before library prep
Insert
Median 300–500 bp
Layer 02 · DNA

Exome or cancer panel

Hybridisation-capture exome across 19,441 genes, 33.9 Mb of target sequence. Where tumour content falls below 50% and cannot be raised by dissection, a cancer panel at higher depth replaces it and the RNA workflow is unchanged.

Exome
19,441 genes · 33.9 Mb target
Coverage
~90× mean on-target
Panel
~500× mean on-target, low tumour content
Reports
Variants and mutational burden
Layer 03 · Derived

Pathway activation

Expression is normalised against reference tissue and resolved into activation levels across annotated signalling pathways. Agents are then ordered by the activation of the pathways they act on.

Output
Activation level per pathway, 0–100
Ranking
Agents ordered by target pathway activity
Basis
Contributing genes and interactions per rank
Cross-check
Whether a variant-bearing gene is expressed

Paired extraction from one dissected region is the point. A variant found in DNA can be checked against whether that gene is transcribed at all, and a pathway running hard without any mutation in it still surfaces. Neither is visible to a DNA panel alone.

Worked case

Interrogate the report

The case above is synthetic and exists to show the format. Ask how the ranking is built, what the level means, or what a tumour board should do with it.

Report assistant
Pick a question, or write your own.
What the assistant will not do

It answers on method and report format only. It will not recommend treatment for a real patient, interpret a real report, or comment on prognosis. Those belong to the treating team and the molecular tumour board.

What you send

Specimen

  • MaterialFFPE block or 10 unstained sections, 10 µm
  • Tumour content≥ 50% in the dissected region
  • PathologyReport confirming diagnosis and site
  • ClinicalPrior lines, current ECOG, relevant comorbidity
  • ConsentPer your institution's molecular testing pathway
What comes back

Report

  • RankingSystemic agents ordered by pathway activation level
  • BasisContributing genes and interactions for each rank
  • VariantsExome or panel findings alongside the expression result
  • QualityCoverage, read depth and RNA integrity for the run
  • FormatPDF for the record, structured export for the board
Scope and limits

Read this before you order

PerMED is decision support for a molecular tumour board. It ranks candidate agents by molecular rationale. It does not predict response, and a high rank is not evidence of benefit in an individual patient.

Ranking is one input among many. Performance status, prior exposure, toxicity, drug access and trial eligibility remain clinical judgements for the treating team.

Evidence base. The ranking method underlying PerMED has been evaluated in peer-reviewed retrospective and prospective work in advanced solid tumours, and presented at international precision-oncology meetings. Those studies were conducted independently of PerMED and are not presented here as PerMED's own results. Full references are supplied to clinical partners on request under confidentiality.

Regulatory status. PerMED reporting is supplied as decision support to the requesting clinician. Classification differs between jurisdictions; we confirm the position for your country in writing before a first case is accepted.

Clinical access

Bring a case to us

Tell us your centre, tumour types and expected volume. We will send the specimen protocol and a sample report.

info@permed.ai