You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: README.md
+24-14Lines changed: 24 additions & 14 deletions
Original file line number
Diff line number
Diff line change
@@ -35,17 +35,22 @@ This is the official codebase of the paper
35
35
36
36
## Contents
37
37
38
-
-[Installation](#installation)
39
-
-[How to prepare data for FlowDock](#how-to-prepare-data-for-flowdock)
40
-
-[How to train FlowDock](#how-to-train-flowdock)
41
-
-[How to evaluate FlowDock](#how-to-evaluate-flowdock)
42
-
-[How to create comparative plots of evaluation results](#how-to-create-comparative-plots-of-evaluation-results)
43
-
-[How to predict new protein-ligand complex structures and their affinities using FlowDock](#how-to-predict-new-protein-ligand-complex-structures-using-flowdock)
44
-
-[For developers](#for-developers)
45
-
-[Docker](#docker)
46
-
-[Acknowledgements](#acknowledgements)
47
-
-[License](#license)
48
-
-[Citing this work](#citing-this-work)
38
+
-[FlowDock](#flowdock)
39
+
-[Description](#description)
40
+
-[Contents](#contents)
41
+
-[Installation](#installation)
42
+
-[How to prepare data for `FlowDock`](#how-to-prepare-data-for-flowdock)
43
+
-[Generating ESM2 embeddings for each protein (optional, cached input data available on SharePoint)](#generating-esm2-embeddings-for-each-protein-optional-cached-input-data-available-on-sharepoint)
44
+
-[Predicting apo protein structures using ESMFold (optional, cached data available on Zenodo)](#predicting-apo-protein-structures-using-esmfold-optional-cached-data-available-on-zenodo)
45
+
-[How to train `FlowDock`](#how-to-train-flowdock)
46
+
-[How to evaluate `FlowDock`](#how-to-evaluate-flowdock)
47
+
-[How to create comparative plots of benchmarking results](#how-to-create-comparative-plots-of-benchmarking-results)
48
+
-[How to predict new protein-ligand complex structures and their affinities using `FlowDock`](#how-to-predict-new-protein-ligand-complex-structures-and-their-affinities-using-flowdock)
If you do not already have a template protein structure available for your target of interest, set `input_template=null` to instead have the sampling script predict the ESMFold structure of your provided `input_protein` sequence before running the sampling pipeline. For more information regarding the input arguments available for sampling, please refer to the config at `configs/sample.yaml`.
361
366
367
+
**NOTE:** To optimize prediction runtimes, a `csv_path` can be specified instead of the `input_receptor`, `input_ligand`, and `input_template` CLI arguments to perform *batched* prediction for a collection of protein-ligand sequence pairs, each represented as a CSV row containing column values for `id`, `input_receptor`, `input_ligand`, and `input_template`. Additionally, disabling `visualize_sample_trajectories` may reduce storage requirements when predicting a large batch of inputs.
368
+
369
+
For instance, one can perform batched prediction as follows:
0 commit comments