greenbeing-binary

This is a binary classification version of the finetuning/evaluation datasets introduced in the freeform text annotations of the greenbeing-proteins dataset.

Proteins from UniProtKB (knowledge base), from select food crops and related species.
Amino acid sequences use IUPAC-IUB codes where letters A-Z map to amino acids.

Tasks

The "query" field currently contains 18 possible labels from the annotation, and the "label" field is the binary class (0 = No, 1 = Yes)

Subcellular Locations

 'cell membrane', # negatives cannot include: cell surface, cell wall
 'endoplasmic reticulum',
 'plastid',
 'chloroplast',
 'vacuole',
 'nucleus',
 'microsome',
 'cytoplasm',
 'extracellular space',
 'mitochondrion',
 'golgi',
 'endosome membrane',
 'cytoskeleton',
 'chromosome',
 'nucleolus',
 'spindle',
 'extracellular matrix',
 'centromere',

All positives are included (between 4 and 673), and ~100 randomly selected annotated proteins without this location are included in negatives.

Dataset builder notebook: https://colab.research.google.com/drive/1QQPjXpqBrd75vP8UbPmZqo-LlcOSyzZu?usp=sharing

Example using a protein embeddings model and SciKit-Learn to build a classifier for "chloroplast" proteins: https://colab.research.google.com/drive/1MCTn8f3oeIKpB6n_8mPumet3ukm7GD8a?usp=sharing

Finetuning split

Reviewed proteins (Swiss-Prot) from above taxa. Each row contains a gene name, species or subspecies, and an amino acid sequence.

A gene name may match multiple entries on UniProt from different accessions.

Evaluation split

Reviewed proteins (Swiss-Prot) from other genera (avocado, carrot, cassava, lychee, prunus (i.e. peaches, cherries, plums)). Each row contains a gene name, species or subspecies, and an amino acid sequence.

A gene name may match multiple entries on UniProt from different accessions.