Skip to main content
Back to challenges
Challenge

Analyze Gene Mutant Dosage and Metastatic Tropism with Scanpy and PyTorch

Mutant allele dosage profoundly alters prognosis and metastatic site preferences in clinical cancer samples. Construct a deep neural net classification pipeline using PyTorch and Scanpy on 60,000 genomic profiles to predict organ-specific metastatic tropism with ROC-AUC greater than 0.85.

Data ScienceHosted by Vera
Sign in free to start
Challenge brief

What you are building

The core problem, expected build, and operating context for this challenge.

Deep learning genomic classification model built with PyTorch and Scanpy to model mutant allele dosage and metastatic tropism.

Delivery guide

How work is evaluated

Evaluation

Evaluates multi-class prediction ROC-AUC for metastatic site prediction based on gene mutation dosage.

Datasets

Challenge datasets

Public datasets, plus the private datasets you have access to.

Evaluation rubric

How submissions are scored

These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.

Dimensions
2 scoring checks
Binary
2 pass or fail dimensions
Ordinal
0 scaled dimensions
Dimension 1

tropism_prediction_validity

Verifies prediction outputs valid organ site.

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Dimension 2

roc_auc

Multi-class ROC-AUC across metastatic organ predictions. • target: 0.85 • range: 0.5-1

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Learning goals

What you should walk away with

  • Process high-dimensional somatic mutation matrices using Scanpy and AnnData formats.

  • Build PyTorch multi-class classification models to predict organ tropism (e.g., liver, brain, bone metastasis).

  • Quantify non-linear interactions between mutant allele copy number (dosage) and metastatic progression.

Resources and assets

Reference links and supporting material

Dataset notes

Dataset derived from 60,000 clinical cancer sample mutation and copy-number profiles with annotated metastatic sites.

Challenge setup

Evaluates multi-class prediction ROC-AUC for metastatic site prediction based on gene mutation dosage.

Challenge input

JSON containing mutated_genes vector and copy_number_alterations vector

PyTorch

Deep learning library for neural network training.

Scanpy

Genomic matrix manipulation library.

Evaluated output

JSON object with predicted_primary_metastatic_site and confidence_distribution

Evaluation checks
  • Verifies prediction outputs valid organ site.
  • Multi-class ROC-AUC across metastatic organ predictions. • target: 0.85 • range: 0.5-1
Configured evaluation evidence
  • Benchmark: OncoTropism-Bench
  • Roc Auc target: 0.85
  • 1 public reference case
Configured runtime
  • Python execution harness
  • Python sandbox (unavailable on Versalist)

Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.

View technical recipe

Configured tools

Action Space
  • Meta · Required
  • Lazy Dev · Optional
  • Azure OpenAI Service · Optional

Evaluation contract

  • tropism_prediction_validity · Weight 1
  • roc_auc · Weight 1

Metrics

  • roc_auc · Target 0.85

Benchmarks

  • OncoTropism-Bench
  • biotech

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON containing mutated_genes vector and copy_number_alterations vector

    Output: JSON object with predicted_primary_metastatic_site and confidence_distribution

    Execution harness: Python

Evaluation tools

  • PyTorch · model_training
  • Scanpy · data_preprocessing

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Analyze Gene Mutant Dosage and Metastatic Tropism with Scanpy and PyTorch