Dawnn model selection and hyperparmeter optimisation

<p>This project is a collection of files to allow users to reproduce the model development and benchmarking in "Dawnn: single-cell differential abundance with neural networks" (Hall and Castellano, under review). Dawnn is a tool for detecting differential abundance in single-cell RNAseq datasets. It is available as an R package <a href="https://github.com/george-hall-ucl/dawnn" target="_blank">here</a>. Please contact us if you are unable to reproduce any of the analysis in our paper.</p> <p>The files in this collection contain to the code to reproduce the model selection and hyperparameter optimisation along with the results.</p> <p><br></p> <p>FILES:</p> <ul> <li><strong>nn_model_choice.py</strong> Python code to select architecture of neural networks and training hyperparameters (learning rate and whether dropout is employed). Uses Keras and Tensorflow. Outputs are stored in <em>model_evaluations_structure_all_nn_results.txt</em>.</li> <li><strong>model_evaluations_structure_all_nn_results.txt</strong> Results from neural network model selection and hyperparameter optimisation in <em>nn_model_choice.py</em>.</li> <li><strong>eval_rf_svm.py</strong> Python code to evaluate the performance of support vector machines and random forests for the same task as neural networks. Results are stored in <em>svm_model_evaluations.txt </em>and<em> rf_model_evaluations.txt</em>, respectively.</li> <li><strong>svm_model_evaluations.txt</strong> Results from evaluation of support vector machines in <em>eval_rf_svm.py</em>.</li> <li><strong>rf_model_evaluations.txt</strong> Results from evaluation of random forests in <em>eval_rf_svm.py</em>.</li> </ul>