PyUUL provides an interface between biological structures and deep learning algorithms

Abstract

Structural bioinformatics suffers from the lack of interfaces connecting biological structures and machine learning methods, making the application of modern neural network architectures impractical. This negatively affects the development of structure-based bioinformatics methods, causing a bottleneck in biological research. Here we present PyUUL (https://pyuul.readthedocs.io/), a library to translate biological structures into 3D tensors, allowing an out-of-the-box application of state-of-the-art deep learning algorithms. The library converts biological macromolecules to data structures typical of computer vision, such as voxels and point clouds, for which extensive machine learning research has been performed. Moreover, PyUUL allows an out-of-the box GPU and sparse calculation. Finally, we demonstrate how PyUUL can be used by researchers to address some typical bioinformatics problems, such as structure recognition and docking.

Intrusion Detection Method Based on Deep Learning

“…In this case, write a timestamp as a Text object during the design and construction of Hadoop-based intrusion detection big data analysis (because only one key is used in the design and construction of Hadoop-based intrusion detection big data analysis) and wrap the timestamp [19] with IntWritable type. In the process of designing and building a Hadoop-based intrusion detection big data analysis, the output record [20] is written only after the feature data is displayed, and its quality code represents the correct feature data store ID. The reduced function is also defined when using Reducer, as defined in the deep learning-based intrusion detection big data store designed and built in this paper as follows: definition of Reducer intrusion detection feature big data access interface of the highest feature data sample [16,17].…”

HPMPdb: A machine learning-ready database of protein molecular phenotypes associated to human missense variants

“…Having in mind that detailed molecular features are likely needed in order to be able to model the protein molecular phenotype caused by each SAV, we ensured that each SAV in HPMPdb is indeed mapped on an experimentally determined protein structure from PDB ( Berman et al., 2000 ). This will allow researchers to include structural information in their tools, favoring the development of novel end-to-end ML pipelines that can natively take protein structures as inputs, as proposed in the recently published ThermoNet ( Bian et al., 2020 ) and in PyUUL ( Orlando et al., 2022 ).…”

“…1 ). In this way, we encourage the users to use the 3D protein structure data available on PDB to contextualize further the possible molecular effects of each SAV, either by extracting features from the corresponding protein structure or by building fully end-to-end ML approaches that take the PDB structure (or a portion centered on the target SAV) and directly use the 3D volume and its content as input for a ML pipeline ( Orlando et al., 2022 ) (e.g. a Neural Network).…”

“…a Neural Network). A similar approach has been adopted to predict protein thermal stability ( Bian et al., 2020 ) and in the PyUUL library ( Orlando et al., 2022 ).…”

Is theoretical chemistry entering a new era of big data learning and reasoning: deep electron cloud-activity and field-activity relationships

“…Fortunately, deep learning (with multihidden layer neural networks) has provided us with a powerful tool for learning massive molecular entities 20,21 . Deep learning has achieved unprecedented success in many fields of artificial intelligence, especially image analysis and classification of 3D point clouds 22,23 , where the learning ability of big data and its advantages in capturing local details using convolutional techniques have been brilliant.…”