mogam-ai/taxonomy-embeddings

📊 NCBI Dataset

This dataset is derived from the NCBI and was incorporated durgin the pretraining process with CDS-BART. It contains randonmly selected 500 mRNA sequences from each of the four taxonomies: bacteria, invertebrate, plant, and fungi, totaling 2000 sequences.

⁉️ Dataset Contents

• Sequence: The mRNA sequences corresponding to each of the four taxonomies
• Label: The labels representing the four different taxonomies: bacteria, invertebrate, plant, and fungi [0,1,2,3]

🎯 Purpose

This dataset is intended for extracting embeddings using a large-language model The primary purpose of the dataset is to evaluate models trained with distinct patterns associated with different taxonomies. CDS-BART available at GitHub

🔧Usage

from datasets import load_dataset
from huggingface_hub import hf_hub_download

dataset = load_dataset("mogam-ai/taxonomy-embeddings", data_files = 'CDSBART_BEST_3_df.csv')
embedding = torch.load(hf_hub_download(repo_id="mogam-ai/taxonomy-embeddings", 
                        filename="emb_no_eos_CDSBART_BEST_3.pt", repo_type="dataset"))