PathoMapPathoMap
1 / 10

01Purpose

PathoMap, a simple automated tool for genomic annotation.

A researcher starts with a gene and a variant, and gets back one row of 104 columns drawn from eleven public sources: clinical, functional, conservation, pathway, expression and model-organism data. Every value records where it came from.

gene,variant HBB,p.Glu6Val TP53,p.R175H clin_siggnomAD SIFTphyloP ReactomeHPA Bgee… Pathogenic1.2% deleterious0.60 10 pathsenriched enriched104 model-organism columns residueconservation source one group per species p.Glu7 conserved recorded

Tasneem Mohamed · Mohamed Reyad
Pre-seed · 2026

02Problem

One variant takes eleven databases and a hand-built spreadsheet.

Today getting there means chaining VEP, bcftools and gnomAD, reshaping the output in R, and writing a converter for each file format.

1 variant p.Glu6Val Clinical ClinVar · gnomAD Functional VEP · CADD · SIFT Conservation phyloP · GERP Context Reactome · KEGG · GO · HPA · Bgee R + shell scripts a converter per format then a spreadsheet no record of versions species lookup: a separate step outside the pipeline 11 sources, 4 groups, 1 manual join
Eleven is the actual integration list, grouped here so it reads. The join is the part nobody automates.

03Solution

One command returns that row, including the columns labs build by hand.

One command joins all eleven sources, instead of a script per format. Every value carries its source, so the row rebuilds. The species columns ship in the row, instead of a side lookup.

gene,variant HBB,p.Glu6Val TP53,p.R175H one command genechrom clin_siggnomAD SIFTphyloP ReactomeHPA Bgee HBB11 Pathogenic1.2% deleterious0.60 10 pathsenrichedenriched model-organism columns residueconservation source one group per species p.Glu7 conserved recorded 104 columns. Every value carries its source, so any number can be traced back.

04Product

It resolves the variant first, which is where most errors begin.

Four steps, one command, and a run log that records what succeeded, failed or was served from cache.

Why nowVariant volumes outgrew manual review. Public APIs fail often enough to need a fallback. Model-organism data is finally reachable through stable interfaces.

01

Resolve

Notation to GRCh38 coordinates, checked against ClinVar.
SPDI: chromosome, position, bases.

02

Annotate

Clinical, frequency and protein-level predictions.
VEP: Ensembl's effect predictor.

03

Orthologs

Same gene in other species, residue projected.
Residue: the amino acid position.

04

Report

One table, plus a log of what each source returned.

104

columns, each source-tagged

660

tests, all offline

11

data sources

~1 min

per variant

05Customer discovery

Discovery is in progress: three segments, three things we are testing.

N of 25 interviews done. Themes and quotes go in the appendix as they close, and we are not quoting findings we have not coded.

Genomics cores

Run annotations for other labs.

Site licence before seats.

Rare-disease labs

Need model-organism context for follow-up work.

Cross-species is the first step dropped.

Model-organism groups

Already work this way, slowly.

Reproducibility beats coverage.

06Competition

Cross-species tools exist. They are not batch pipelines.

MARRVEL does the lookup one query at a time. VEP and ANNOVAR batch-annotate thousands of variants, but stay inside the human genome.

interactive lookup → batch pipeline human only → model organisms VarSome Franklin VEP ANNOVAR MARRVEL in-house scripts PathoMap batch and cross-species

Placement follows each tool's documented use, not a benchmark. MARRVEL is the neighbour to benchmark against.

The gap is workflow, not capability: a list of a thousand variants, one table, every value traceable.

07Market

A $150M segment today, heading for $500M by 2030.

Annotation software for research. The larger interpretation market above it is mostly clinical; it is context, and we do not claim it today.

Annotation segment, today Same segment, 2030 $150M $500M Published estimates. The interpretation market above this segment is $470–570M today, mostly clinical: context, not our target.

08Business model

Labs pay per seat, cores pay for a licence, teaching is free.

Private deployment for institutions that cannot send data out. The source is proprietary; the method and the operational work are the hard parts to copy.

Labs

Per-seat subscription, billed annually.

Pilot pricing with first cores

Cores

One site licence covering every user at the facility.

Set with first customers

Institutions

Private deployment behind their firewall.

Scoped per deployment

09Team

Two builders, and a gap we know about.

We built the pipeline, the data integration, the interface and the deployment. What we do not have in-house is clinical and regulatory experience, or anyone whose job is selling.

Tasneem Mohamed

Co-founder

Mohamed Reyad

Co-founder

Supervisor: Olaitan I. Awe, Ph.D. (ASBCB)
Acknowledgements: Hackbio, ASBCB, Ensembl, NCBI, UCSC

10The ask

Raise to reach paying pilots and a published benchmark.

Twelve months of runway and three checkable milestones, building this as a company rather than a side project. Amount and terms to discuss.

Q1

Pilot

First core running paid workloads

Q2

Benchmark

Published accuracy and runtime numbers

Q3

Licences

Site licences and private deployments

Q4

Company

Incorporation and first commercial hire