Skip to main content
Back to challenges
Challenge

Intrinsically Disordered Region (IDR) Demethylase Target Predictor with BioEmbeddings

Intrinsically disordered regions (IDRs) restrain genomic targeting of histone and RNA demethylases across species. You will build a sequence feature extraction and protein targeting model using BioEmbeddings and PyTorch to predict IDR-mediated targeting inhibition with 88% cross-validation accuracy.

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.

Train a protein sequence classification model using BioEmbeddings representations to identify IDR-driven demethylase genomic target restraints.

Delivery guide

How work is evaluated

Evaluation

Evaluates protein IDR target restraint classification accuracy against experimentally validated demethylase mutants.

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

idr_detection_validity

Ensure sequence with high disorder yields idr_present=True

Binary check

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

Dimension 2

accuracy

Cross-validation classification accuracy • target: 0.88 • range: 0-1

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

  • Extract per-residue protein embeddings using ProtT5 and ESM-2 models via BioEmbeddings

  • Identify intrinsically disordered region boundaries and biophysical property metrics

  • Predict demethylase genomic localization restraint changes upon IDR deletion

Resources and assets

Reference links and supporting material

Dataset notes

Protein sequence dataset containing 1,200 plant and mammalian demethylase variants with annotated IDR boundaries.

Challenge setup

Evaluates protein IDR target restraint classification accuracy against experimentally validated demethylase mutants.

Challenge input

JSON object with protein_id and fasta_sequence string

BioEmbeddings

Utility for extracting embeddings from protein transformers.

PyTorch

Neural network modeling engine.

Evaluated output

JSON object with idr_present boolean and restraint_score float

Evaluation checks
  • Ensure sequence with high disorder yields idr_present=True
  • Cross-validation classification accuracy • target: 0.88 • range: 0-1
Configured evaluation evidence
  • Benchmark: ProteinLM-Bench
  • Accuracy target: 0.88
  • 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
  • Langbase · Required
  • Gemma 2 · Optional
  • AWS · Optional

Evaluation contract

  • idr_detection_validity · Weight 1
  • accuracy · Weight 1

Metrics

  • accuracy · Target 0.88

Benchmarks

  • ProteinLM-Bench
  • biotech

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON object with protein_id and fasta_sequence string

    Output: JSON object with idr_present boolean and restraint_score float

    Execution harness: Python

Evaluation tools

  • BioEmbeddings · Protein Representation Toolkit
  • PyTorch · Deep Learning Engine

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Intrinsically Disordered Region (IDR) Demethylase Target Predictor with BioEmbeddings