Skip to Main content Skip to Navigation
Master Thesis

Deep learning pour l'analyse génomique des mélanomes canins

Abstract : Mucosal melanoma is a rare and aggressive cancer in humans. Its low prevalence hinders its study and understanding thus requires moving towards an animal model. The dog represents a good model because it naturally develops mucosal melanomas with a high prevalence and strong clinical similarities with humans. The research project I carried out consists of genomic analysis of canine mucosal melanoma with the objective of developing the method to predict the expression of genes involved in tumorigenesis. For this, I set up a bioinformatic tool to predict the expression of genes in humans through a deep learning algorithm. A first result was to transpose this tool to the analysis of the canine genome. Then, I tested and determined the predictive performance of canine gene expression by correlation tests (Pearson r = 0.4). The final step of the project aims to improve the predictive ability of the deep learning models realized through the tool by searching for the optimal hyperparameters of the neural network and widening the learning sample of the model.
Document type :
Master Thesis
Complete list of metadata

Cited literature [20 references]  Display  Hide  Download
Contributor : Agrocampus Ouest <>
Submitted on : Monday, November 18, 2019 - 9:59:33 AM
Last modification on : Monday, July 6, 2020 - 3:38:27 PM


Files produced by the author(s)


  • HAL Id : dumas-02367615, version 1



Camille Kergal. Deep learning pour l'analyse génomique des mélanomes canins. Sciences du Vivant [q-bio]. 2019. ⟨dumas-02367615⟩



Record views


Files downloads