The biomarker intelligence platform
“The literature is mixed” is not an answer
Forty tabs and a spreadsheet will not survive your next big meeting. Motif turns the pile into cited biomarker evidence you can defend.
$6 (one-time) to start. No subscription.
An animation of a Motif run: the query "Find biomarkers for CAR-T resistance in solid tumors" is typed, five steps complete in turn (searching papers, screening them for relevance, extracting biomarker associations, generating a report, and building a knowledge graph), and the panel resolves into a graph linking CAR-T resistance to PD-1, CTLA-4, TILs, CD8+ T cells, T cell exhaustion and an epigenetic barrier score.
So the names actually match
- PubMed(opens in new window)
- PubMed Central(opens in new window)
- Europe PMC(opens in new window)
- CIViC(opens in new window)
- DGIdb(opens in new window)
- ClinVar(opens in new window)
- ClinPGx(opens in new window)
- gnomAD(opens in new window)
- FDA(opens in new window)
- ClinicalTrials.gov(opens in new window)
- Gene Ontology(opens in new window)
- Reactome(opens in new window)
- HGNC(opens in new window)
- Ensembl(opens in new window)
- TCGA(opens in new window)
- NCBI Gene(opens in new window)
53
databases, so names match
Full text
tables and figures, not abstracts
Effect sizes
the numbers the room will ask for
Cited
every finding back to its paper
01
Question
Ask what the room will ask
02
Table
Associations with effect sizes
03
Grade
Why the claim holds
04
Report
Cited, export-ready
05
Graph
How it connects next
How it works
The room will ask how you compared them
Tabs are impressions. Motif extracts the finding so you can filter it, grade it, and cite it.
Extraction
n=12 and a trial look the same in a tab
One paper is mice. One used a different assay. Motif pulls effect size, interval, p-value, sample size, and design from full text, tables, and figures.
- Effect sizes, intervals, p-values, and sample sizes when the paper reports them
- Tied back to the source paper and PMID
- Full text, tables, and figures. Not abstracts.
Associations, showing three extracted findings. PD-L1 TPS at or above 50 percent predicts improved progression-free survival on pembrolizumab versus chemotherapy in untreated non-small-cell lung cancer, hazard ratio 0.50 with a 95 percent confidence interval of 0.37 to 0.68, p below 0.001, n=305, from a randomised controlled trial, graded High certainty. Neurofilament light chain in cerebrospinal fluid is associated with faster motor progression in early Parkinson's disease, beta 0.31 per standard deviation, graded Moderate. KRAS G12C co-mutation confers resistance to erlotinib in EGFR-mutant lung adenocarcinoma, hazard ratio 1.74, graded Low. Each finding carries the paper it came from and its PubMed identifier.
Cross-referencing
HER2, ERBB2, and neu are not three markers
Each entity is resolved against 53 databases. A name in a paper becomes an accession you can look up. A mismatch becomes visible.
- Genes, proteins, diseases, drugs, pathways, and related biomarker types
- Linked to UniProt, MONDO, ClinVar, CIViC, gnomAD, Reactome, and more
- Accessions and definitions travel with the finding
A cross-reference panel for the gene ERBB2, also known as HER2, matched against four of Motif's 53 reference databases. UniProt gives accession P04626, reviewed in Swiss-Prot, organism Homo sapiens. ClinVar reports expert-panel review status with a pathogenic classification for ERBB2 amplification. CIViC records evidence level B, predictive significance, in breast carcinoma. The FDA record lists trastuzumab, indicated for HER2-positive breast cancer, approved in 1998.
Your graph
A folder of PDFs does not connect
A list of 200 papers tells you what exists. A graph tells you how it connects. Each paper extends yours.
- Relationships such as inhibits, activates, associated with, and predictive of
- Filter by context, species, study design, or certainty
- Export from BibTeX to Neo4j
A biomarker knowledge graph centred on EGFR in lung cancer. EGFR is associated with lung cancer and predicts sensitivity to erlotinib, while KRAS confers resistance to it. TP53, PD-L1 and IL-6 are each associated with lung cancer, EGFR upregulates PD-L1, and miR-21 correlates with EGFR.
Evidence you can defend
A case report is not a 10,000-patient trial
PubMed treats them the same. Motif grades the evidence so you can say why a claim holds.
An evidence certainty panel adapted from GRADE. The finding is Moderate certainty, with study design, sample size, effect size and risk-of-bias summary shown so a reader can see why the grade landed where it did.
A discordant evidence panel for tumour mutational burden as a predictor of overall survival. One study of 810 patients reports a hazard ratio of 0.61 with a 95 percent confidence interval of 0.44 to 0.85. Another study of 442 patients reports a hazard ratio of 1.28 with a confidence interval of 0.98 to 1.67. Motif classifies the statistical conflict as incompatible, because the confidence intervals exclude one another, and resolves it as context-dependent: the effect differs by disease context, melanoma versus colorectal cancer.
When papers disagree, that is the finding. Motif shows whether it is a real contradiction or a difference in context.
Who Motif is for
The same meeting. Three versions.
Shortlist before the indication is locked
A pile of PDFs is not a table you can defend in the room.
It worked in one cohort and vanished
See which findings hold across contexts, and which only worked once.
You are building the evidence package
Study design, n, and assay have to travel with the claim.
Pricing
Start for $6, once
No trial clock. Run the question before the meeting and decide afterwards.
Starter
$6 (one-time)
One question before the meeting
Pro
$30
Regular diligence, every week
Max
$100
Large programs, many questions
Enterprise
Custom
Your org graph, shared
Walk in with an answer
One query. Structured, graded, and cited.