{"id":28562204,"url":"https://github.com/borgwardtlab/rnaglib-experiments","last_synced_at":"2026-03-07T19:31:22.335Z","repository":{"id":293554967,"uuid":"801502638","full_name":"BorgwardtLab/rnaglib-experiments","owner":"BorgwardtLab","description":null,"archived":false,"fork":false,"pushed_at":"2025-09-23T07:34:47.000Z","size":19914,"stargazers_count":1,"open_issues_count":0,"forks_count":0,"subscribers_count":2,"default_branch":"main","last_synced_at":"2025-09-23T09:24:23.994Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":null,"language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/BorgwardtLab.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2024-05-16T11:02:19.000Z","updated_at":"2025-09-23T07:34:50.000Z","dependencies_parsed_at":null,"dependency_job_id":"99638c9c-a858-4a3e-b408-dcb0f180091b","html_url":"https://github.com/BorgwardtLab/rnaglib-experiments","commit_stats":null,"previous_names":["borgwardtlab/rnaglib-experiments"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/BorgwardtLab/rnaglib-experiments","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/BorgwardtLab%2Frnaglib-experiments","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/BorgwardtLab%2Frnaglib-experiments/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/BorgwardtLab%2Frnaglib-experiments/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/BorgwardtLab%2Frnaglib-experiments/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/BorgwardtLab","download_url":"https://codeload.github.com/BorgwardtLab/rnaglib-experiments/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/BorgwardtLab%2Frnaglib-experiments/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":30227785,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-03-07T19:01:10.287Z","status":"ssl_error","status_checked_at":"2026-03-07T18:59:58.103Z","response_time":53,"last_error":"SSL_read: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":[],"created_at":"2025-06-10T12:02:09.851Z","updated_at":"2026-03-07T19:31:22.215Z","avatar_url":"https://github.com/BorgwardtLab.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# rnaglib-experiments\n\nThis repository contains all code used for our experiments conducted with the `rnaglib` task suite. For full documentation please refer to [rnaglib.org](rnaglib.org) and [REDACTED].\n\n# Reproducing experiments\n\nThis repository provides the necessary code to reproduce the three main experiments reported in our preprint: number of layers ablation study, splitting strategy ablation study and representation ablation study\n\n* Prior to running any of the three experiments detailed below, you need to create the datasets. To do so, please run `python create_datasets.py`. This will create the datasets relevant to each of the tasks and store them in a folder named `roots`\n\n* To reproduce the number of layers ablation study, run `python run_exp_nb_layers.py`. Once this is done, JSON files containing the training data will be stored in `results`. Then, you will be able to run `python plotting_scripts/make_plot_nb_layers.py` which will create (if default parameters are kept) a file named `plotting_scripts/nb_layers_ablation_2.5D.pdf` corresponding to Figure 4a of our preprint. Please note that you can tune some parameters, for instance `representation` in order to visualize the impact of the number of layers when using a different representation than 2.5D. In this case, you need to change accordingly the parameters in `run_exp_nb_layers.py` and in `make_plot_nb_layers.py`\n\n* To reproduce the splitting strategy ablation study, run `python run_exp_splitting.py`, which will train models and dump the relevant JSONs in `results`. In order to reproduce the associated plot, run `python plotting_scripts/make_plot_splitting.py`. This will create a file named `plotting_scripts/splitting_ablation.pdf` reproducing Figure 3a of our preprint.\n\n* To reproduce the representation ablation study, run `python run_exp_representations.py`, which will train models and dump the relevant JSONs in `results`. In order to reproduce the associated plot, run `python plotting_scripts/make_plot_representation.py`. This will create a file named `plotting_scripts/representation_ablation.pdf` reproducing Figure 4b of our preprint.\n\n* To reproduce the benchmark table , run `python run_exp_splitting.py`, which will train each model with its default splitting and 2.5D representations with its best hyperparameters. Then run `python plotting_scripts/make_table_benchmark.py`. This will create a file named `plotting_scripts/final_benchmark.pdf` reproducing Table 2 of our preprint, alongside a CSV file `plotting_scripts/final_benchmark.csv`.\n\nOnce a training has been made in specific conditions, it won't be re-run if a subsequently used script needs to run it, unless you change the `retrain` parameter to `True`. Therefore, running experiments which trainings partially overlap won't lead to a waste of time.\n\n* To reproduce the timing results, run `python timing.py`\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fborgwardtlab%2Frnaglib-experiments","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fborgwardtlab%2Frnaglib-experiments","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fborgwardtlab%2Frnaglib-experiments/lists"}