Modular Multi–Source Prediction of Drug Side–Effects With DruGNN

Bongini, Pietro; Pancino, Niccolò; Dimitri, Giovanna Maria; Bianchini, Monica; Scarselli, Franco; Lió, Píetro

doi:10.1109/tcbb.2022.3175362

Cited by 17 publications

(20 citation statements)

References 37 publications

(49 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Furthermore, the AI era opens new perspectives to drug design [31] and protein-ligand simulations [32]. The possibility of very fast tailoring of ligands for Gly-induced pockets through molecular graph generation with graph neural networks is just a very recent example of AI-related advancement [33], whose products can also be automatically evaluated with an AI predictor of side-effects [34]. Thus, our structural Bioinformatics survey on the human pathogenic variant database suggests a fast rational selection of those Mendelian disorders where X/Gly-mutations determine damages that can be repaired simply by swallowing a pill: a very powerful shortcut to restrict all the experimental procedures that are needed to achieve disease remediation.…”

Section: Discussionmentioning

confidence: 99%

Some Mendelian Disorders could be fixed with a pill? A Structural Bioinformatic investigation

Bongini

Iovinelli

Trezza

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

It has been recently suggested that amino acid replacements with Gly can modify the shape of protein surfaces and, hence, protein dynamics and functions. We have browsed ClinVar, the database of all the reported variants of clinical relevance, to identify all the proteins having missense X/Gly mutations that determine Mendelian disorders. We have found 959 benign and 875 pathogenic X/Gly substitutions. Pathogenicity origins were initially searched in the distribution profiles of replaced amino acids. These profiles indicate that Mendelian disorders including Gly-replacements arise mainly from substitutions of amino acids bearing bulky hydrophobic side chains, thus reducing protein core stability. In the case mutated proteins were structurally defined, we could give a deeper insight into pathogenicity mechanisms, checking whether Gly-mutations altered protein shapes, modifying water surface dynamics and, hence, the physiological protein-protein interaction processes. In several cases, indeed, we have found that pathological Gly-mutants present additional surface pockets, suggesting that the new pockets could be the target of a pharmacological strategy for Mendelian disorder remediation.HighlightsClinVar has been scanned to find signals for pathogenicity due to X/Gly mutationsPathogenicity origins of X/Gly replacements have been structurally analyzedX/Gly mutations can create protein surface pockets with binding capabilitiesGly-formed new protein binding sites can be the target for Mendelian disorder curesAI procedures will expand the search for structural damages due to X/Gly mutationsGraphical abstract

show abstract

Section: Discussionmentioning

confidence: 99%

Some Mendelian Disorders could be fixed with a pill? A Structural Bioinformatic investigation

Bongini

Iovinelli

Trezza

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Especially graph neural networks (GNNs) [ 61 ]. Moreover, graph-based models are used heavily in the biological domain to predict the properties of new compounds, estimate their activity levels, predict their side effects [ 13 ], and generate candidate molecular structures [ 54 ]. However, due to their irregular nature, graphs are inherently hard to compute, becoming a challenging task [ 10 , 23 , 45 , 46 , 51 , 66 ].…”

Section: Introductionmentioning

confidence: 99%

Distributed large-scale graph processing on FPGAs

et al. 2023

View full text Add to dashboard Cite

Processing large-scale graphs is challenging due to the nature of the computation that causes irregular memory access patterns. Managing such irregular accesses may cause significant performance degradation on both CPUs and GPUs. Thus, recent research trends propose graph processing acceleration with Field-Programmable Gate Arrays (FPGA). FPGAs are programmable hardware devices that can be fully customised to perform specific tasks in a highly parallel and efficient manner. However, FPGAs have a limited amount of on-chip memory that cannot fit the entire graph. Due to the limited device memory size, data needs to be repeatedly transferred to and from the FPGA on-chip memory, which makes data transfer time dominate over the computation time. A possible way to overcome the FPGA accelerators’ resource limitation is to engage a multi-FPGA distributed architecture and use an efficient partitioning scheme. Such a scheme aims to increase data locality and minimise communication between different partitions. This work proposes an FPGA processing engine that overlaps, hides and customises all data transfers so that the FPGA accelerator is fully utilised. This engine is integrated into a framework for using FPGA clusters and is able to use an offline partitioning method to facilitate the distribution of large-scale graphs. The proposed framework uses Hadoop at a higher level to map a graph to the underlying hardware platform. The higher layer of computation is responsible for gathering the blocks of data that have been pre-processed and stored on the host’s file system and distribute to a lower layer of computation made of FPGAs. We show how graph partitioning combined with an FPGA architecture will lead to high performance, even when the graph has Millions of vertices and Billions of edges. In the case of the PageRank algorithm, widely used for ranking the importance of nodes in a graph, compared to state-of-the-art CPU and GPU solutions, our implementation is the fastest, achieving a speedup of 13 compared to 8 and 3 respectively. Moreover, in the case of the large-scale graphs, the GPU solution fails due to memory limitations while the CPU solution achieves a speedup of 12 compared to the 26x achieved by our FPGA solution. Other state-of-the-art FPGA solutions are 28 times slower than our proposed solution. When the size of a graph limits the performance of a single FPGA device, our performance model shows that using multi-FPGAs in a distributed system can further improve the performance by about 12x. This highlights our implementation efficiency for large datasets not fitting in the on-chip memory of a hardware device.

show abstract

“…Only recently, a method named DruGNN was proposed to overcome the preprocessing limit by using graph neural networks (GNNs) [14], although it still relies on preprocessed MFs. GNNs are powerful connectionist models for graph-structured data processing, which have become practical tools for any problem involving graphs, thanks to their capability of processing relational data directly in graph form and calculating an output at each node or edge, with minimal loss of information [15].…”

Section: Introductionmentioning

confidence: 99%

“…Following the approach proposed in DruGNN [14], the data of interest consist of a heterogeneous graph including two types of nodes (drugs and genes) and three types of edges (drug-gene, drug-drug, and gene-gene relationships). Formally, the task is a node-focused, multi-class, multi-label classification problem: namely, the GNN model is trained to predict the DSEs associated with the drug nodes.…”

Section: Introductionmentioning

confidence: 99%

“…In this case, since the NF is produced by a ML model, it is dynamically adapted during the learning procedure for the DSE prediction task to maximize the performance of the general model. • DruGNN: The model is similar the one introduced in [14]. This network is a recurrent GNN model [16] which exploits an inductive-transductive learning scheme to predict DSEs on drug nodes.…”

mentioning

confidence: 99%

See 1 more Smart Citation

Drug Side Effect Prediction with Deep Learning Molecular Embedding in a Graph-of-Graphs Domain

et al. 2022

Self Cite

View full text Add to dashboard Cite

Drug side effects (DSEs), or adverse drug reactions (ADRs), constitute an important health risk, given the approximately 197,000 annual DSE deaths in Europe alone. Therefore, during the drug development process, DSE detection is of utmost importance, and the occurrence of ADRs prevents many candidate molecules from going through clinical trials. Thus, early prediction of DSEs has the potential to massively reduce drug development times and costs. In this work, data are represented in a non-euclidean manner, in the form of a graph-of-graphs domain. In such a domain, structures of molecule are represented by molecular graphs, each of which becomes a node in the higher-level graph. In the latter, nodes stand for drugs and genes, and arcs represent their relationships. This relational nature represents an important novelty for the DSE prediction task, and it is directly used during the prediction. For this purpose, the MolecularGNN model is proposed. This new classifier is based on graph neural networks, a connectionist model capable of processing data in the form of graphs. The approach represents an improvement over a previous method, called DruGNN, as it is also capable of extracting information from the graph-based molecular structures, producing a task-based neural fingerprint (NF) of the molecule which is adapted to the specific task. The architecture has been compared with other GNN models in terms of performance, showing that the proposed approach is very promising.

show abstract

Modular Multi–Source Prediction of Drug Side–Effects With DruGNN

Cited by 17 publications

References 37 publications

Some Mendelian Disorders could be fixed with a pill? A Structural Bioinformatic investigation

Some Mendelian Disorders could be fixed with a pill? A Structural Bioinformatic investigation

Distributed large-scale graph processing on FPGAs

Drug Side Effect Prediction with Deep Learning Molecular Embedding in a Graph-of-Graphs Domain

Contact Info

Product

Resources

About