R

Rinalmo

Developed by multimolecule
RiNALMo is a non-coding RNA (ncRNA) model pre-trained based on the masked language modeling (MLM) objective, trained through self-supervised learning on a large number of non-coding RNA sequences.
Downloads 21.38k
Release Time : 9/10/2024

Model Overview

RiNALMo is a BERT-style model specifically designed for processing non-coding RNA sequences, pre-trained via masked language modeling tasks, and can be used for RNA sequence analysis and structure prediction.

Model Features

Large-scale pre-training
Pre-trained on 36 million unique ncRNA sequences, covering multiple RNA databases.
Self-supervised learning
Trained via masked language modeling tasks without the need for manually labeled data.
Sequence diversity handling
Uses MMSeqs2 to cluster sequences, ensuring diversity in training batches.
High-performance architecture
Adopts a 33-layer Transformer architecture with a hidden layer size of 1280 and 20 heads, suitable for processing long sequences.

Model Capabilities

RNA sequence analysis
RNA structure prediction
Masked nucleotide prediction
RNA sequence feature extraction

Use Cases

Bioinformatics
HIV-1 RNA analysis
Predicting masked nucleotides in HIV-1 RNA sequences
The model can accurately predict the most likely nucleotides at masked positions
microRNA analysis
Predicting masked nucleotides in microRNA-21 sequences
The model can identify key nucleotides in microRNA sequences
RNA research
Non-coding RNA function prediction
Predicting the function of non-coding RNAs through sequence features
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
© 2025AIbase