---
title: Oncology Analytics FAQ
description: Frequently asked questions about oncology analytics tools on the platform.
tags:
  - oncology
  - faq
---

## Oncology Analytics FAQ

These answers cover the [oncology analytics](/docs/oncology) biomarkers and their interpretation — how TMB and MSI-H differ from raw mutation counts, HLA typing and peptide lengths for neoantigen prediction, and the research-use-only scope of these tools.

## How is TMB different from the total mutation count?

TMB (tumor mutational burden) normalizes the mutation count by the size of the sequenced region, expressed as mutations per megabase (mut/Mb). This normalization allows comparison across different sequencing panels. A tumor with 150 mutations detected across a 1.5 Mb panel has a TMB of 100 mut/Mb, while 150 mutations across a 50 Mb exome would be 3 mut/Mb. Only non-synonymous coding mutations are typically counted. The normalization is critical because panel size varies widely between targeted panels (0.5-10 Mb), whole exome (30-60 Mb), and whole genome sequencing.

## Can I use targeted panel data for TMB calculation?

Yes, but specify the `panel_size_mb` parameter so the tool can normalize correctly. Panel-based TMB estimates have higher variance than whole-exome estimates, especially for panels smaller than 1 Mb. For panels 1-2 Mb, TMB estimates correlate well with WES-derived TMB (R > 0.7). For panels smaller than 0.5 Mb, TMB estimates may be unreliable and should be interpreted with caution.

## What is the difference between MSI-H and TMB-high?

MSI-H (microsatellite instability-high) and TMB-high are related but distinct biomarkers. MSI-H tumors often have high TMB because mismatch repair deficiency causes both microsatellite instability and elevated mutation rates. However, TMB-high tumors are not always MSI-H. For example, melanomas and NSCLC can be TMB-high due to UV or smoking-related mutagenesis without mismatch repair deficiency. Both biomarkers independently predict response to immune checkpoint inhibitors, and both have FDA-approved indications for pembrolizumab.

## Do I need to know the patient's HLA type for neoantigen prediction?

Yes. HLA typing is essential for neoantigen prediction because neoantigens must bind to the patient's specific MHC molecules to be presented to T cells. You should provide the patient's HLA class I alleles (typically HLA-A, HLA-B, HLA-C, two alleles each for six total). If HLA typing is unavailable, you can use the most common alleles in the patient's ethnic background as an approximation, but this reduces prediction accuracy.

## What peptide lengths should I use for MHC binding prediction?

For MHC class I, use 8-11mer peptides, with 9mers being the most common binding length for most HLA-A and HLA-B alleles. For MHC class II, use 13-25mer peptides, with 15mers being the standard. The neoantigen prediction tool automatically generates peptides of appropriate lengths from somatic mutations, so you typically only need to specify lengths when using the MHC binding tool directly.

## How are neoantigen candidates ranked?

Candidates are ranked by a composite score that weighs multiple factors: MHC binding affinity (IC50 < 50 nM strongly preferred), gene expression level (higher TPM increases antigen presentation likelihood), clonality (clonal mutations present in all tumor cells are preferred), and agretopicity (the ratio of wildtype to mutant binding affinity, where higher ratios indicate more tumor-specific presentation). The top-ranked candidates represent the best potential targets for personalized cancer vaccines or adoptive T cell therapy.

## Can these tools be used for clinical decision-making?

These tools are intended for research use only (RUO). While the algorithms and thresholds are based on published clinical evidence, the results should not be used as the sole basis for clinical decisions without validation by CLIA-certified laboratory testing and clinical review. The FDA-approved companion diagnostics (e.g., Myriad myChoice CDx for HRD, FoundationOne CDx for TMB) use validated laboratory processes that differ from research-grade analysis.

## Read next