| |
|
|
| Registros recuperados : 298 | |
| 8. |  | CIAPPESONI, G. Mejora genética ovina en Uruguay: lecciones aprendidas. In: Seminario Internacional "Mejoramiento genético y competitividad del rubro ovino en la región de la Araucanía", Temuco, Chile, 18 de marzo 2016.[Presentación Oral] 61 diap. En el marco del Programa de Trazabilidad y Mejoramiento Genético Ovino.| Biblioteca(s): INIA Las Brujas. |
|    |
| 14. | | CIAPPESONI, G.; GOLDBERG, V. Genetic parameters for body weight, worm resistance, packed cell volume and FAMACHA@ under natural infection in Corriedale sheep. Volume Electronic Poster Session - Species - Ovine, 351. In: Proceedings of the World Congress on Genetics Applied to Livestock Production, 11., Aotea Centre Auckland, New Zealand: WCGALP, ICAR, 11-16 feb 2018, 351.| Biblioteca(s): INIA Las Brujas. |
| |
| 16. | | ZOPPOLO, R.; CIAPPESONI, G. Introducción. Tecnologías con enfoque en la productividad y la calidad. Tercera sección. Tecnologías con enfoque en la productividad y la calidad. Editores: Roberto Zoppolo y Gabriel Ciappesoni. In: Georgina Paula García-Inza; José María Paruelo; Roberto Zoppolo. (eds). Aportes científicos y tecnológicos del Instituto Nacional de Investigación Agropecuaria (INIA) del Uruguay a las trayectorias agroecológicas. Ciudad Autónoma de Buenos Aires : Fundación CICCUS, 2023. p.295-300. p.295-300.| Biblioteca(s): INIA Las Brujas. |
| |
| Registros recuperados : 298 | |
|
|
|
|
Registro completo
|
|
Biblioteca (s) : |
INIA Las Brujas. |
|
Fecha actual : |
17/11/2023 |
|
Actualizado : |
17/11/2023 |
|
Tipo de producción científica : |
Poster |
|
Autor : |
FARIELLO, M.I.; ARBOLEYA, L.; BELZARENA, D.; DE LOS SANTOS, L.; ELENTER, J.; ETCHEBARNE, G.; HOUNIE, I.; CIAPPESONI, G.; NAVAJAS, E.; LECUMBERRY, F. |
|
Afiliación : |
MARIA INÉS FARIELLO, Facultad de Ingeniería, Universidad de la República, Uruguay; Institut Pasteur de Montevideo, Uruguay; LUCÍA ARBOLEYA, Facultad de Ingeniería, Universidad de la República, Uruguay; DIEGO BELZARENA, Facultad de Ingeniería, Universidad de la República, Uruguay; LEONARDO DE LOS SANTOS, Facultad de Ingeniería, Universidad de la República, Uruguay; JUAN ELENTER, Facultad de Ingeniería, Universidad de la República, Uruguay; GUILLERMO ETCHEBARNE, Facultad de Ingeniería, Universidad de la República, Uruguay; IGNACIO HOUNIE, Facultad de Ingeniería, Universidad de la República, Uruguay; CARLOS GABRIEL CIAPPESONI SCARONE, INIA (Instituto Nacional de Investigación Agropecuaria), Uruguay; ELLY ANA NAVAJAS VALENTINI, INIA (Instituto Nacional de Investigación Agropecuaria), Uruguay; FEDERICO LECUMBERRY, Facultad de Ingeniería, Universidad de la República, Uruguay; Institut Pasteur de Montevideo, Uruguay. |
|
Título : |
Something old, something new, something borrowed : Evaluation of different neural network architectures for genomic prediction. [poster] |
|
Fecha de publicación : |
2023 |
|
Fuente / Imprenta : |
En: Plant & Animal Genome Conference : PAG 30, San Diego, California, USA, 13-18 january 2023. |
|
Descripción física : |
Editorial: Plant and Animal Genome Conference (PAG). |
|
Idioma : |
Inglés |
|
Notas : |
Este trabajo fue parcialmente financiado por la Universidad de la República y el proyecto ANII FDA 1_2018_1_154364. -- LICENCIA: Licencia Creative Commons Atribución - No Comercial - Sin Derivadas (CC - By-NC-ND 4.0). |
|
Contenido : |
ABSTRACT.- Genome enabled prediction of complex traits aims to predict a measurable characteristic of an organism using their genetic information. In the present work we address diverse traits and organisms including Yeast growth, Wheat yield, Jersey bull fertility and various Holstein cattle milk-related traits. We benchmark several popular Machine Learning models: Bayesian and penalized linear regressions, kernel methods, and Decision Tree ensembles. Through exhaustive hyperparameter tuning we outperform state-of-the-art results in most datasets. We also evaluate two codification techniques for input data and perform ablation studies to assess robustness to genetic markers - i.e input features- elimination. We also explore different Deep Learning architectures for this task. We propose and evaluate Convolutional Neural Network (CNN) architectures, showing that using residual connections improves performance but that in some cases Fully Connected Networks outperform CNNs. We link this to the fact that absolute positions are relevant in genomes, and thus, CNN's translational equivariance may not be an adequate inductive bias for tackling this problem. We evaluate Graph Neural Network (GNN) architectures by formulating trait prediction as a node regression problem on a population graph, where each node represents an individual, and edges association between their genetic information. We evaluate the transferability of these graphical models and find that the extent to which they exploit neighborhood information is limited. By combining CNN and GNN architectures, we could outperform all other models for predicting milk yield in Holstein cattle.The methods that are based on neural networks can be computationally demanding when used on high density chips or sequence data, even more when fully connected layers are used. To overcome this problem, we propose to obtain a new representation of the input vector by using the intermediate representation (code) of an Autoencoder (AE). Currently we are evaluating the performance benchmarks. Another common issue when using these databases is the missing data or the combination of chips of different SNP's numbers. Again, we propose to use AE for imputing the missing values. One of the main focuses of this work was to explore the feasibility of employing modern deep learning architectures in Genomic Prediction. In this regard, it was possible to train highly over-parameterized architectures and still obtain good generalization. For some datasets and traits, these models outperform all others. However, this did not hold for all the models, traits and datasets studied. Besides, whether the gains in performance outweigh the increase in model size and thus its training and inference computational cost, and lack of interpretability, calls for further discussion. MenosABSTRACT.- Genome enabled prediction of complex traits aims to predict a measurable characteristic of an organism using their genetic information. In the present work we address diverse traits and organisms including Yeast growth, Wheat yield, Jersey bull fertility and various Holstein cattle milk-related traits. We benchmark several popular Machine Learning models: Bayesian and penalized linear regressions, kernel methods, and Decision Tree ensembles. Through exhaustive hyperparameter tuning we outperform state-of-the-art results in most datasets. We also evaluate two codification techniques for input data and perform ablation studies to assess robustness to genetic markers - i.e input features- elimination. We also explore different Deep Learning architectures for this task. We propose and evaluate Convolutional Neural Network (CNN) architectures, showing that using residual connections improves performance but that in some cases Fully Connected Networks outperform CNNs. We link this to the fact that absolute positions are relevant in genomes, and thus, CNN's translational equivariance may not be an adequate inductive bias for tackling this problem. We evaluate Graph Neural Network (GNN) architectures by formulating trait prediction as a node regression problem on a population graph, where each node represents an individual, and edges association between their genetic information. We evaluate the transferability of these graphical models and find that the extent to which t... Presentar Todo |
|
Palabras claves : |
Deep learning; Predicción genómica; Signal processing. |
|
Asunto categoría : |
L10 Genética y mejoramiento animal |
|
URL : |
http://www.ainfo.inia.uy/digital/bitstream/item/17417/1/Fariello-PAG-2023-FABDEEHCNL23.pdf
|
|
Marc : |
LEADER 03979nam a2200277 a 4500
001 1064372
005 2023-11-17
008 2023 bl uuuu u00u1 u #d
100 1 $aFARIELLO, M.I.
245 $aSomething old, something new, something borrowed$bEvaluation of different neural network architectures for genomic prediction. [poster]$h[electronic resource]
260 $aEn: Plant & Animal Genome Conference : PAG 30, San Diego, California, USA, 13-18 january 2023.$c2023
300 $cEditorial: Plant and Animal Genome Conference (PAG).
500 $aEste trabajo fue parcialmente financiado por la Universidad de la República y el proyecto ANII FDA 1_2018_1_154364. -- LICENCIA: Licencia Creative Commons Atribución - No Comercial - Sin Derivadas (CC - By-NC-ND 4.0).
520 $aABSTRACT.- Genome enabled prediction of complex traits aims to predict a measurable characteristic of an organism using their genetic information. In the present work we address diverse traits and organisms including Yeast growth, Wheat yield, Jersey bull fertility and various Holstein cattle milk-related traits. We benchmark several popular Machine Learning models: Bayesian and penalized linear regressions, kernel methods, and Decision Tree ensembles. Through exhaustive hyperparameter tuning we outperform state-of-the-art results in most datasets. We also evaluate two codification techniques for input data and perform ablation studies to assess robustness to genetic markers - i.e input features- elimination. We also explore different Deep Learning architectures for this task. We propose and evaluate Convolutional Neural Network (CNN) architectures, showing that using residual connections improves performance but that in some cases Fully Connected Networks outperform CNNs. We link this to the fact that absolute positions are relevant in genomes, and thus, CNN's translational equivariance may not be an adequate inductive bias for tackling this problem. We evaluate Graph Neural Network (GNN) architectures by formulating trait prediction as a node regression problem on a population graph, where each node represents an individual, and edges association between their genetic information. We evaluate the transferability of these graphical models and find that the extent to which they exploit neighborhood information is limited. By combining CNN and GNN architectures, we could outperform all other models for predicting milk yield in Holstein cattle.The methods that are based on neural networks can be computationally demanding when used on high density chips or sequence data, even more when fully connected layers are used. To overcome this problem, we propose to obtain a new representation of the input vector by using the intermediate representation (code) of an Autoencoder (AE). Currently we are evaluating the performance benchmarks. Another common issue when using these databases is the missing data or the combination of chips of different SNP's numbers. Again, we propose to use AE for imputing the missing values. One of the main focuses of this work was to explore the feasibility of employing modern deep learning architectures in Genomic Prediction. In this regard, it was possible to train highly over-parameterized architectures and still obtain good generalization. For some datasets and traits, these models outperform all others. However, this did not hold for all the models, traits and datasets studied. Besides, whether the gains in performance outweigh the increase in model size and thus its training and inference computational cost, and lack of interpretability, calls for further discussion.
653 $aDeep learning
653 $aPredicción genómica
653 $aSignal processing
700 1 $aARBOLEYA, L.
700 1 $aBELZARENA, D.
700 1 $aDE LOS SANTOS, L.
700 1 $aELENTER, J.
700 1 $aETCHEBARNE, G.
700 1 $aHOUNIE, I.
700 1 $aCIAPPESONI, G.
700 1 $aNAVAJAS, E.
700 1 $aLECUMBERRY, F.
Descargar
Esconder MarcPresentar Marc Completo |
|
Registro original : |
INIA Las Brujas (LB) |
|
|
Biblioteca
|
Identificación
|
Origen
|
Tipo / Formato
|
Clasificación
|
Cutter
|
Registro
|
Volumen
|
Estado
|
Volver
|
| Expresión de búsqueda válido. Check! |
|
|
