README.md

metadata

Author: hy3074
Date: 25.11.24
Description: Synthetic DNA-to-AA ORF Translation Dataset
tags:
  - dna
  - protein
  - bioinformatics
  - synthetic-biology
  - sequence-to-sequence
  - translation
tasks:
  - sequence-to-sequence
language:
  - en
dataset_info:
  features:
    - name: src
      dtype: string
    - name: tgt
      dtype: string
  splits:
    - name: train
      num_bytes: 3332329
      num_examples: 4900
    - name: validation
      num_bytes: 476047
      num_examples: 700
    - name: test
      num_bytes: 952094
      num_examples: 1400
  download_size: 2371456
  dataset_size: 4760470
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: validation
        path: data/validation-*
      - split: test
        path: data/test-*

Synthetic DNA → Amino Acid ORF Translation Dataset

🧬 Summary

This dataset contains synthetic DNA sequences (src) and their corresponding amino acid sequences (tgt), where the label is defined as the first valid ORF from the 5' end:

From the first ATG (start codon) encountered in the sequence
to the first in-frame stop codon (TAA, TAG, or TGA) that follows it,
translated into amino acids (excluding the stop).

The dataset is designed for training and evaluating models that learn DNA→protein translation rules and ORF detection in a slightly non-trivial, noisy setting.

---

📁 Dataset Structure

Data fields

Each example is a JSON object with the following fields:

• src (string):
  A DNA sequence (characters A, C, G, T only).

  • Length is variable, from a minimum (e.g. 6 nt) up to a specified maximum.
  • The sequence is mostly random, but guaranteed to contain at least one valid ORF.

• tgt (string):
  The amino acid sequence translated from the ORF defined as:

  • The first ATG found when scanning from the 5' end.
  • The first in-frame stop codon (TAA, TAG, TGA) after that ATG.
  • Translation uses the standard genetic code.
  • The stop codon is not included in the amino acid string.

Example (conceptual):

{
  "src": "ACGTTGATGAAAACCGGGTAGTTCG...",
  "tgt": "MKTVG"
}