Figure - PMC

Skip to main content

An official website of the United States government

Here's how you know

Here's how you know

Official websites use .gov
A .gov website belongs to an official government organization in the United States.

Secure .gov websites use HTTPS
A lock ( ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

View full-text article in PMC

. 2020 Jul 11;48(14):7690–7699. doi: 10.1093/nar/gkaa583

Search in PMC
Search in PubMed
View in NLM Catalog
Add to search

© The Author(s) 2020. Published by Oxford University Press on behalf of Nucleic Acids Research.

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com

PMC Copyright notice

Figure 2. — Outline of the RNAmigos pipeline. A base pairing network is passed as input to RNAmigos. In training mode, it is paired with a native ligand (Target) from which a target fingerprint y is constructed. The embedding network (RGCN) produces a matrix of node embeddings of dimension n × d where n is the number of nodes in the graph, and d is a fixed embedding size. This is followed by a pooling step which reduces node embeddings to a single graph-level vector. Finally, the graph representation is fed through a multi-layer perceptron (MLP) to produce a predicted fingerprint that minimizes the distance to the native fingerprint y. The fingerprint is then used to search for similar ligands to the prediction in a ligand screen and thus enriches the probability of identifying an active compound. The RGCN network is pre-trained using an unsupervised node embedding framework which allows us to leverage structural patterns from a large dataset of RNA structures. This network is trained to generate embeddings which minimize the distance () between kernel similarities k(u, ) and embedding similarities 〈_u, 〉.

Inline graphic — Outline of the RNAmigos pipeline. A base pairing network is passed as input to RNAmigos. In training mode, it is paired with a native ligand (Target) from which a target fingerprint y is constructed. The embedding network (RGCN) produces a matrix of node embeddings of dimension n × d where n is the number of nodes in the graph, and d is a fixed embedding size. This is followed by a pooling step which reduces node embeddings to a single graph-level vector. Finally, the graph representation is fed through a multi-layer perceptron (MLP) to produce a predicted fingerprint that minimizes the distance to the native fingerprint y. The fingerprint is then used to search for similar ligands to the prediction in a ligand screen and thus enriches the probability of identifying an active compound. The RGCN network is pre-trained using an unsupervised node embedding framework which allows us to leverage structural patterns from a large dataset of RNA structures. This network is trained to generate embeddings which minimize the distance () between kernel similarities k(u, ) and embedding similarities 〈_u, 〉.