Understanding RNA Regulation with AI
Challenges in RNA Data
Despite having a lot of genomic data, we still need to understand the RNA regulatory code better. Current genomic models use techniques from other fields but lack biological insights. Experimental methods to study RNA are often costly and time-consuming. Machine learning on genetic sequences offers a more efficient and affordable solution to predict important cellular processes like RNA degradation and alternative splicing.
Introducing Self-Supervised Learning
Recent studies suggest using foundation models in genomics that employ self-supervised learning (SSL) on unlabeled data. These models aim to perform well across various tasks, even with fewer labeled samples. However, genomic sequences can be complex due to low diversity, which can lead to ineffective RNA predictions.
Meet Orthrus: A New RNA Model
Researchers from the Vector Institute and the University of Toronto have developed Orthrus, an RNA foundation model. Orthrus uses contrastive learning with biological data to improve RNA predictions. It focuses on maximizing similarities between RNA transcripts from closely related species, using data from over 400 mammalian species. This model significantly outperforms existing genomic models, especially in low-data situations.
How Orthrus Works
Orthrus uses contrastive learning to analyze RNA splicing and functional similarities. It pairs RNA isoforms and orthologous sequences to train the model effectively. The Mamba encoder processes RNA data, and the model is evaluated on various RNA tasks like predicting RNA half-life and gene classification.
Benefits of Orthrus
Orthrus builds structured representations of RNA transcripts, enhancing the understanding of functionally related sequences. The model has shown superior performance in predicting RNA properties, making it highly effective in low-data environments.
Future of RNA Predictions
Orthrus captures RNA diversity by modeling evolutionary relationships, leading to strong predictions even when data is limited. While it excels in many areas, it may face challenges in cases where variations have minimal impact on RNA properties.
Stay Connected
Explore the Paper, Model on HF, and GitHub for more details. Follow us on Twitter, join our Telegram Channel, and LinkedIn Group. If you enjoy our work, sign up for our newsletter and join our 50k+ ML SubReddit.
Join Our Upcoming Webinar
Join us on Oct 29, 2024, to learn about the best platform for serving fine-tuned models: the Predibase Inference Engine.
Transform Your Business with AI
To stay competitive, utilize Orthrus to enhance your operations. Identify automation opportunities, set measurable KPIs, choose suitable AI solutions, and implement gradually. For AI KPI management advice, connect with us at hello@itinai.com. For ongoing insights, follow us on Telegram or Twitter.
Revolutionize Your Sales and Customer Engagement
Discover how AI can transform your sales processes and customer interactions at itinai.com.
