file structure

proteinshake/readme.md

@@ -0,0 +1,15 @@
+Filestructure:
+
+**backend:** Collects and processes raw pdb files from various databases. Creates Collections.
+    **adapters:** Database adapters to PDB, AFDB, etc.
+    **protein:** The protein object specification.
+    **collection:** A set of protein objects.
+    **database:*** Unifying database mirror that allows query access to create a collection.
+**frontend:** High level specifications for datasets and tasks.
+    **datasets:** Load a collection and apply transforms.
+    **tasks:** Various ML prediction problems. Takes a dataset and rearranges the data (sklearn-style X_train, y_train, ...) such that it can be fed to the model (X) and evaluated (y).
+    **evaluators:** Groups relevant metrics for a given class of problems (e.g. classification or regression).
+    **splitters:** Takes a dataset and generates split indices, either from the data/labels itself or from other resources.
+**datasets:** Some relevant collections of proteins, for preprocessed hosting. Each dataset covers some biological topic.
+**tasks:** The actually implemented biological tasks for the end-user, consisting of a frontend.task in combination with a specific dataset.
+**transforms:** Various functions to transform proteins, representations, and frameworks.