Malignant cell detection
Malignant cell detection turns Computational Biology single-cell CNV output into oncology-facing tumor biology summaries. It does not recompute copy-number signal. Instead, it consumes the CompBio single_cell_cnv handoff or result JSON and organizes malignant labels, CNV events, subclones, mutation overlap, and response features for oncology workflows.
Prerequisites
- A completed CompBio
single_cell_cnvrun withoncology_handoff.jsonor equivalent result JSON - Optional mutation landscape result JSON for gene overlap with inferred CNV events
- Optional communication or survival result JSON for downstream context
Launching the pipeline
Open the study Analyze page and use Malignant cell detection, selecting the completed CompBio CNV run (and optionally a mutation landscape run) to pull context from.
Key parameters
The Analyze page's dropdowns send run/experiment IDs, not paths — the backend resolves each artifact path itself from the ID. Direct *_result_path parameters below still work (for API use), but are not what the UI sends.
| Parameter | Required | Purpose |
|---|---|---|
compbio_run_id / compbio_experiment_id | One of these, or cnv_result_path | Completed CompBio single-cell CNV run, picked from the Analyze page's dropdowns |
mutation_run_id | No | Completed oncology mutation landscape run, for inferred event-gene overlap |
communication_run_id | No | Completed oncology communication run, for context-aware handoff notes |
survival_run_id | No | Completed oncology survival run, for outcome context |
cnv_result_path | No (API-only alternative to compbio_run_id) | CompBio single-cell CNV handoff or result JSON path |
mutation_result_path / communication_result_path / survival_result_path | No (API-only alternatives to the *_run_id params) | Direct result JSON paths |
Outputs
| Output | Description |
|---|---|
summary | Source run, malignant fraction, subclone count, event count, and linked-context flags |
malignant_populations | Groups called likely malignant with fraction, burden, confidence, reason, and top CNV evidence |
subclone_tme_features | CNV subclone rows prepared for communication sender grouping |
cnv_events | Inferred top events carried over from the CompBio result |
mutation_overlap | Overlap between inferred event genes and supplied mutation/CNA result genes |
survival_response_features | Feature rows for malignant fraction, CNV burden, subclone fractions, and key inferred events |
workflow_handoffs | Suggested metadata columns and groupings for communication, mutation, survival, and immune workflows |
caveats | Research-use warnings from the CompBio run plus oncology interpretation caveats |
The frontend result panel shows malignant populations, top CNV evidence, subclone TME handoff rows, mutation overlap, downstream feature counts, and download buttons for the summary, survival-feature, and malignant-population artifacts.
Typical use cases
- Review malignant clusters as tumor biology before running TME communication.
- Stratify communication by malignant subclone sender and immune/stromal receiver.
- Compare inferred CNV-event genes with mutation/CNA outputs.
- Export malignant fraction, CNV burden, and subclone fractions as response or survival features.
- Ground multi-run interpretation in CNV, immune, mutation, and outcome evidence.
Caveats
This analysis is research-only. It depends on the quality of the CompBio CNV run, especially gene order and reference-cell selection. Use DNA copy-number, mutation, histology, or orthogonal annotations where available for confirmation.