Reference Guide

← Back to tool

How to Read the Results

A brief guide to the statistical values displayed by the Tumor Likelihood Tool and how to interpret them.

Statistic

Prior Probability

The baseline frequency of a tumor type in the GENIE dataset, before any mutation or CNA evidence is considered. It reflects how common each cancer type is among sequenced samples.

Example
Lung Prior 15.13%
Thyroid Prior 1.46%

Lung cancer is more common in GENIE, so it starts with a higher prior.

Statistic

Posterior Probability

The updated probability of a tumor type after incorporating your mutation and CNA evidence. This is the main result — tumor types are ranked by posterior, shown as the bold percentage on each card.

Example — BRAF V600E entered
Skin & Melanoma 46.48%
Thyroid 22.44%

Thyroid jumped from 1.46% prior to 22.44% posterior because BRAF V600E is strongly enriched there.

posterior ∝ prior × evidence

Evidence

Fold Enrichment

How many times more (or fewer) an alteration occurs in a tumor type compared to the overall dataset average. A value of 5.7x means the mutation is 5.7 times more frequent in that tumor type than expected.

BRAF V600E frequency by tumor type
Thyroid 13.24x enriched
Skin & Melanoma 5.71x enriched
Lung 0.38x depleted

Values >1 are enriched (green bar). Values <1 are depleted (orange bar) — the mutation occurs less than expected.

Evidence

Affected Count / Group Total

The raw frequency displayed as affected / total (percentage). Affected is the number of samples with that alteration in the tumor type. Total is the number of panel-covered samples in that tumor type.

BRAF V600E in Skin & Melanoma
1,531 / 7,727 19.8%

Of 7,727 melanoma samples whose panel covers BRAF, 1,531 carry V600E. The denominator only counts samples whose sequencing panel covers the gene (panel-aware analysis).

Statistic

Log Bayes Factor

The natural logarithm of the Bayes factor — a measure of how strongly one alteration shifts the probability toward (or away from) a tumor type. This is the engine behind the posterior update.

Interpretation scale
log BF > 2 Strong evidence for
log BF ≈ 0 Uninformative
log BF < −2 Strong evidence against

Each evidence term contributes a weighted log BF. The sum of all weighted log BFs plus the log prior gives the final log score.

Evidence

Evidence Types & Weights

Each alteration you enter generates one or more evidence terms. Different types carry different default weights reflecting their discriminatory power.

Default weights
Allele-level mutation (e.g. BRAF V600E) 1.2x
Copy number alteration (e.g. ERBB2 amp) 0.5x
Gene-level mutation (fallback if allele not found) 0.1x

Allele-specific mutations are the most informative, so they receive the highest weight. If a specific protein change isn't in the database, the tool falls back to gene-level evidence at a lower weight.

Interface

Drill-Down into Subtypes

Each result card represents a broad tumor group (e.g. LUNG). Click on it to expand and see the detailed cancer subtypes within that group (e.g. Lung Adenocarcinoma, Lung Squamous Cell Carcinoma), along with subtype-specific alteration frequencies.

How it works

The broad tumor type ranking uses enrichment statistics computed across 23 major groups. When you expand a card, you see the CANCER_TYPE_DETAILED subtypes within that group, ranked by their own posterior probability. This lets you distinguish, for example, whether a lung sample is more likely adenocarcinoma or squamous cell.

Reading a Result Card

Here's how to interpret each piece of information on a tumor-type result card.

  1. Tumor type name and posterior percentage appear at the top. This is the probability that the sample belongs to this tumor type, given the alterations you entered.
  2. Prior shows the baseline prevalence. Compare it to the posterior to see how much the evidence shifted the probability. A large jump means strong evidence.
  3. Frequency bars show each alteration's frequency in that tumor type. Green bars indicate enrichment (>1x), orange bars indicate depletion (<1x). The label shows the fold enrichment and the raw count.
  4. Click the card to expand detailed subtypes. Each subtype shows its own posterior and per-alteration frequencies, letting you narrow down the specific cancer type.