Read tumor and normal data before choosing targets
A sequencing result becomes a candidate source only after its identity, tumor specificity and biological support are checked. This is step two of the vaccine guide.
Before you start: DNA, RNA and protein describe different layers. A neoantigen arises from altered sequence but still needs processing and recognition evidence.
Where this step sits
This step supplies the source list. Human leukocyte antigen (HLA) matching and target selection come afterward. A larger source list can include more real candidates and more uncertain calls.
The intuition: compare before calling a difference
A laboratory reads many short pieces of sequence. Variant calling is the process of inferring an alteration from those observations. The program compares reads with a reference and evaluates support, quality and competing explanations. It does not observe a perfectly complete tumor genome directly.
Matched normal material helps distinguish inherited differences from tumor-associated changes. The normal specimen’s suitability and contamination still need review. Tumor-only analysis estimates some of these distinctions, but cannot always establish them confidently.
What the measurements contribute
| Layer | Useful question | Limit to record |
|---|---|---|
| Tumor DNA | Is the sequence alteration supported? | Coverage, artifacts, tumor fraction and copy number |
| Matched normal DNA | Is the alteration absent from the appropriate normal comparison? | Sampling and detection limits |
| Tumor RNA | Is the altered transcript supported? | Mixed-cell composition, expression depth and processing |
| Protein or peptide evidence | Does a downstream product exist? | Method coverage, cell origin and detection limits |
A targeted panel reads selected regions. Whole-exome sequencing focuses largely on coding regions. Whole-genome sequencing covers a broader range of sequence. Greater breadth does not by itself guarantee adequate depth, accurate structural calls or a useful antigen.
Allele fraction is not tumor coverage
Variant allele fraction is the proportion of reads at a locus supporting the variant. Normal-cell admixture, copy-number changes and the number of mutant copies affect it. It is not automatically the percentage of cancer cells carrying a target.
Use copies, alleles and clones for a worked interpretation. A truncal alteration is shared broadly across a tumor’s evolutionary branches; a subclonal alteration occurs in a subset. One small biopsy may not establish either pattern across every site.
Worked example: maintain an evidence ledger
These are invented candidates, not patient data or a scoring algorithm.
| Candidate | Evidence | Next unresolved question |
|---|---|---|
| A | Tumor DNA call, absent in adequately covered normal, mutant RNA observed | Processing and display |
| B | Tumor DNA call, normal coverage too low | Tumor specificity |
| C | RNA junction only, with weak read support | Reproducibility and artifact control |
| D | Strong DNA support, no RNA detection in a shallow assay | Whether non-detection is informative |
Candidate A has progressed further on the ledger. None has yet demonstrated human leukocyte antigen (HLA) presentation or T-cell recognition. Candidate D should not be called definitively unexpressed without adequate sampling and sensitivity.
What can go wrong at this step
Fixation artifacts, misaligned reads, contamination and uncertain junctions can create misleading candidates. Sampling can miss a real alteration. A reference transcript may also be the wrong transcript for translating the altered sequence. Preserve methods, versions, coverage and manual decisions so the candidate can be traced to its source.
Try it
A variant appears in half the reads. Can it be called present in half the cancer cells?
Answer: No. Purity, copy number and mutant-copy multiplicity are needed. Read fraction and cancer-cell fraction are different quantities.
Explain it back
“Matched normal addresses ___; RNA addresses ___; neither alone proves ___.” One answer: “tumor specificity; expression; surface presentation or useful recognition.”
Takeaway
Source quality, expression and tumor coverage are separate entries in a candidate’s evidence ledger.
Next: HLA matching and presentation.
Sources and scope
Source check: October 8, 2026; expert and learner review pending. This is a reading framework, not a laboratory calling protocol.
- NHGRI DNA sequencing overview.
- Primary neoantigen prediction and validation study.
- Copies, alleles and clones — sourced quantitative examples.