Molecular Tumor Board · Internal Demo

From sequencer to
treatment decision

How our NGS tertiary-analysis pipeline turns a TSO500 panel into an ESMO-2024–compliant report the board can act on — in one automated run.

What the pipeline does

Input: one TSO500 BAM or pre-called VCF (523 genes, 1.94 Mb)
Calls & annotates somatic variants Mutect2 → Ensembl VEP · HGVS, ClinVar, COSMIC, gnomAD
Adds CNV, fusions, and the three biomarkers TMB · MSI · HRD
Assigns clinical actionability automatically OncoKB + CiVIC → ESCAT tier + AMP/ASCO/CAP oncogenicity
Pulls treatment-focused literature with AMA citations PubMed + Scopus, ranked by therapeutic relevance
Output: a single self-contained HTML report for the board

Eight stages, one command

00 · QCcoverage, purity

→

01 · CallingMutect2

→

02 · AnnotationVEP

→

03 · CNVCNVkit

→

04 · FusionsManta

05 · BiomarkersTMB / MSI / HRD

→

06 · ClinicalOncoKB · CiVIC · ESCAT

→

07 · LiteraturePubMed · Scopus

→

08 · ReportESMO 2024 HTML

targets::tar_make() — fully orchestrated, reproducible, cached
VCF-only input skips BAM-dependent stages (CNV, fusions, MSI, HRD) automatically

How we use it in the tumor board

Step	Who	Action
Before	Bioinformatics	Run pipeline on the case; QC gate must pass (coverage ≥200×, purity ≥10%)
Pre-read	Molecular pathologist	Review HTML report; confirm Tier I/II calls & biomarkers
At board	Oncologist + panel	Discuss actionable alterations top-down by ESCAT tier
Decision	Board consensus	Match to on-label drug, trial, or off-label use; cite the report's literature
After	Coordinator	Archive report (de-identified); log decision & rationale

One report drives the discussion — no manual variant look-ups during the meeting.

What each result means for the board

Result	Report shows	Board uses it to…
Somatic variants	Ranked table + VAF, oncogenicity	Identify the driver(s)
ESCAT tier	I → X per alteration	Prioritise what is actionable today
TMB	mut/Mb + High/Low	Consider immunotherapy (≥10)
MSI	MSS / MSI-H	Consider pembrolizumab if MSI-H
HRD	score + status	Consider PARP inhibitor if positive
Literature	AMA-cited evidence	Justify the treatment choice

Sample report · mock case

Patient & sample

Case ID: DEMO-OV-001
Diagnosis: HG serous ovarian
Specimen: FFPE tumour
Panel: TSO500 · GRCh38
Mean coverage: 612× PASS
Tumour purity: 45%

Biomarkers

TMB: 4.6 mut/Mb · Low
MSI: MSS (5.2)
HRD: 58 · Positive

Actionable alterations

Alteration	Tier	Implication
BRCA1 E1836fs	I	PARP inhibitor
BRCA1 deletion	I	Supports HRD
CCNE1 amp	II	Trial eligible
PIK3CA H1047R	III	PI3K pathway
PTEN R130*	III	PI3K pathway
TP53 R248W	X	Not actionable

Board call: BRCA1 LoF + HRD-positive → first-line PARP inhibitor maintenance.

Synthetic / de-identified data for demonstration only — not a real patient.

Reading the ESCAT tiers

Tier	Meaning	Typical board action
I	Ready for routine clinical use	Standard-of-care matched therapy
II	Investigational, strong evidence	Enrol in clinical trial
III	Benefit in other tumour types	Consider off-label / discuss
IV	Preclinical evidence only	Watch; trial if available
X	No actionability / driver context	Document, no action

The report sorts alterations by tier so the board always starts with what matters most.

Takeaway

One run.
One report.
A faster board.

Reproducible, cached, version-controlled — same input, same output
ESMO 2024 compliant · OncoKB + CiVIC evidence built in
Self-contained HTML — opens anywhere, no install for reviewers

Live demo report: htlin222.github.io/ngs-tertiary-analysis-skills

Not a prototype — already running

62-patient pan-cancer cohort

Run profile

Patients (consecutive): 62
Monthly batches: 8
Cancer types: 11
Panel: TSO500 · 523 genes
Caller: DRAGEN v2.6
Stage: advanced / late-stage
Completed end-to-end: 62 / 62

Cancer-type mix

Type	n	Type	n
Endometrial	16	NSCLC	8
Breast	14	Ovarian	7
Colorectal	9	Others	8

Single-center Taiwanese tertiary cancer center · prospective enrolment.

Aggregate, de-identified — every case ran through the same pipeline shown above.

Cohort results — the numbers

Biomarkers & therapy match

Metric	n / 62	%
On-label match (L1/2/3A)	21	33.9
Off-label match (L3B/4)	38	61.3
TMB-High (≥10 mut/Mb)	8	12.9
HRD-positive (GIS ≥42)	6	9.7
Resistance flag (R1/R2)	6	9.7
MSI-High	2	3.2

≈1 in 3 advanced tumors had an on-label match — in line with international cohorts (20–40%).

Most recurrent oncogenic drivers

Gene	Patients
TP53	26
KRAS	16
PIK3CA	16
PTEN	11
ARID1A	9
APC · ESR1	7 · 6

Clinical catches

BRCA1 large rearrangement (exon 7 loss, FC 0.15) → on-label PARPi in HRD-positive breast
KRAS/NRAS resistance variants flagged anti-EGFR ineligibility in 6 colorectal cases

Patient reports openly browseable · numbers verified from cohort summary (ESMO 2026 submission).

From sequencer totreatment decision