1
|
Brouard C, Mourad R, Vialaneix N. Should we really use graph neural networks for transcriptomic prediction? Brief Bioinform 2024; 25:bbae027. [PMID: 38349060 PMCID: PMC10939369 DOI: 10.1093/bib/bbae027] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 09/27/2023] [Revised: 12/20/2023] [Accepted: 01/17/2024] [Indexed: 02/15/2024] Open
Abstract
The recent development of deep learning methods have undoubtedly led to great improvement in various machine learning tasks, especially in prediction tasks. This type of methods have also been adapted to answer various problems in bioinformatics, including automatic genome annotation, artificial genome generation or phenotype prediction. In particular, a specific type of deep learning method, called graph neural network (GNN) has repeatedly been reported as a good candidate to predict phenotypes from gene expression because its ability to embed information on gene regulation or co-expression through the use of a gene network. However, up to date, no complete and reproducible benchmark has ever been performed to analyze the trade-off between cost and benefit of this approach compared to more standard (and simpler) machine learning methods. In this article, we provide such a benchmark, based on clear and comparable policies to evaluate the different methods on several datasets. Our conclusion is that GNN rarely provides a real improvement in prediction performance, especially when compared to the computation effort required by the methods. Our findings on a limited but controlled simulated dataset shows that this could be explained by the limited quality or predictive power of the input biological gene network itself.
Collapse
Affiliation(s)
- Céline Brouard
- Université Fédérale de Toulouse, INRAE, MIAT, 31326 Castanet-Tolosan, France
| | - Raphaël Mourad
- Université Fédérale de Toulouse, INRAE, MIAT, 31326 Castanet-Tolosan, France
- Université Paul Sabatier, 31062 Toulouse, France
| | - Nathalie Vialaneix
- Université Fédérale de Toulouse, INRAE, MIAT, 31326 Castanet-Tolosan, France
| |
Collapse
|
2
|
Pomiès L, Brouard C, Duruflé H, Maigné É, Carré C, Gody L, Trösser F, Katsirelos G, Mangin B, Langlade NB, de Givry S. Gene regulatory network inference methodology for genomic and transcriptomic data acquired in genetically related heterozygote individuals. Bioinformatics 2022; 38:4127-4134. [PMID: 35792837 DOI: 10.1093/bioinformatics/btac445] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 10/28/2021] [Revised: 06/17/2022] [Accepted: 07/05/2022] [Indexed: 12/24/2022] Open
Abstract
MOTIVATION Inferring gene regulatory networks in non-independent genetically related panels is a methodological challenge. This hampers evolutionary and biological studies using heterozygote individuals such as in wild sunflower populations or cultivated hybrids. RESULTS First, we simulated 100 datasets of gene expressions and polymorphisms, displaying the same gene expression distributions, heterozygosities and heritabilities as in our dataset including 173 genes and 353 genotypes measured in sunflower hybrids. Secondly, we performed a meta-analysis based on six inference methods [least absolute shrinkage and selection operator (Lasso), Random Forests, Bayesian Networks, Markov Random Fields, Ordinary Least Square and fast inference of networks from directed regulation (Findr)] and selected the minimal density networks for better accuracy with 64 edges connecting 79 genes and 0.35 area under precision and recall (AUPR) score on average. We identified that triangles and mutual edges are prone to errors in the inferred networks. Applied on classical datasets without heterozygotes, our strategy produced a 0.65 AUPR score for one dataset of the DREAM5 Systems Genetics Challenge. Finally, we applied our method to an experimental dataset from sunflower hybrids. We successfully inferred a network composed of 105 genes connected by 106 putative regulations with a major connected component. AVAILABILITY AND IMPLEMENTATION Our inference methodology dedicated to genomic and transcriptomic data is available at https://forgemia.inra.fr/sunrise/inference_methods. SUPPLEMENTARY INFORMATION Supplementary data are available at Bioinformatics online.
Collapse
Affiliation(s)
- Lise Pomiès
- MIAT, Université Fédérale de Toulouse, INRAE, Castanet-Tolosan 31326, France
| | - Céline Brouard
- MIAT, Université Fédérale de Toulouse, INRAE, Castanet-Tolosan 31326, France
| | - Harold Duruflé
- LIPME, Université de Toulouse, INRAE, CNRS, Castanet-Tolosan 31326, France
| | - Élise Maigné
- MIAT, Université Fédérale de Toulouse, INRAE, Castanet-Tolosan 31326, France
| | - Clément Carré
- MIAT, Université Fédérale de Toulouse, INRAE, Castanet-Tolosan 31326, France
| | - Louise Gody
- LIPME, Université de Toulouse, INRAE, CNRS, Castanet-Tolosan 31326, France
| | - Fulya Trösser
- MIAT, Université Fédérale de Toulouse, INRAE, Castanet-Tolosan 31326, France
| | - George Katsirelos
- MIA-Paris, AgroParisTech, Université Paris-Saclay, INRAE, Paris 75231, France
| | - Brigitte Mangin
- LIPME, Université de Toulouse, INRAE, CNRS, Castanet-Tolosan 31326, France
| | - Nicolas B Langlade
- LIPME, Université de Toulouse, INRAE, CNRS, Castanet-Tolosan 31326, France
| | - Simon de Givry
- MIAT, Université Fédérale de Toulouse, INRAE, Castanet-Tolosan 31326, France
| |
Collapse
|