4a70f0ef93
* Docs: Installation FAQ space and minor RFD3 docs updates Installation FAQ: created a document to specify any common installation issues and questions. Should be continuously updated based on logged issues and questions. Not specific to any model. RFD3: - changed the checkpoint files specified in the examples to rfd3_latest.ckpt - updated information in input.md to clarify information based on recent issues that had been submitted * Docs: Symlinks for RF3 and MPNN docs, RFD3 README minor edits RF3 and MPNN: folders, index files, and symlinks were created in order to provide space for eventual RF3 and MPNN docs. Several small changes in the RFD3 README to improve readability and add a pointer to the PPI tutorial as a starting point for someone new to RFdiffusion tools. * First draft of enzyme design tutorial. Minor typo fixes in other documents. * First draft of nucleic acid binder tutorial, minor edits to the other tutorials * Completed enzyme design tutorial, removal of NA binder tutorial from index Made changes based on edits from Saman, added images, and created zip file containing sample outputs for an enzyme design tutorial. I am waiting on edits for the NA binder design tutorial, so for now I have removed it from the documentation index. * Removing file related to in-progress NA binder tutorial * Removing file related to in-progress NA binder tutorial * Update ppi_design_tutorial.md - Added information about useful CLI arguments - Cleaned up the introduction - Added section for what one might do with the designs from RFD3 - Added a note about hotspot residues also being in the `contig` (information from Rafi's TTT talk) - Fixed minor sphinx heading issue * Reorganizing RFD3 documentation Reorganized files into an `examples` and a `tutorials` folders to clean up the RFD3 docs folder and align its organization with the RF3 docs folder. Any edits made in the files are related to changing the paths to reflect these changes. * Docs: Designability vs. Diversity document Created a document describing the settings that can impact the designability and diversity of structures output by RFdiffusion3, the information is based on the talk Rafi gave at Tech Tea Time in January. * Minor grammar fixes in designability vs diversity document * Update models/rfd3/docs/tutorials/ppi_design_tutorial.md Co-authored-by: Rafael Brent <105883594+RafiBrent@users.noreply.github.com> * Update models/rfd3/docs/tutorials/enzyme_design_tutorial.md Co-authored-by: Rafael Brent <105883594+RafiBrent@users.noreply.github.com> * Update models/rfd3/docs/designability_vs_diversity.md Co-authored-by: Rafael Brent <105883594+RafiBrent@users.noreply.github.com> * Update models/rfd3/docs/designability_vs_diversity.md Co-authored-by: Rafael Brent <105883594+RafiBrent@users.noreply.github.com> --------- Co-authored-by: Jasper Butcher <66851659+Ubiquinone-dot@users.noreply.github.com> Co-authored-by: Rafael Brent <105883594+RafiBrent@users.noreply.github.com>
3.5 KiB
Designability vs. Diversity
When using RFdiffusion3 there is a balance between designability and diversity of generated structures. Increasing the diversity of the designs will lead to a greater number of novel folds, however, there will also be a larger portion of structures that have low confidence scores when refolded.
Whether you are struggling to produce designable structures or you are looking to increase the diversity of the folds you see, here are a few settings to try changing:
-
Low temperature sampling:
One can increase
inference_sampler.step_scaleand decreaseinference_sampler.gamma_0to decrease the sampling space that RFdiffusion3 has access to, similar to what lowering the temperature does in physics-based design methods. These settings directly change how the RFdiffusion3 inference engine works, so these options are specified in the CLI, and are not options you specify in your input JSON or YAML file.Here are what these settings do:
inference_sampler.step_scale: Changing this value (default 1.5) changes the diffusion step size, or how much you go towards the most probable result. Increasing this setting will increase the designability of the output structures, as these are more probable, but will also decrease the diversity of the produced structures.inference_sampler.gamma_0: Changing this value (default 0.6) will change how much noise is added at each step in the inference trajectory. Decreasing this setting will increase the designability of the output structures as the reduced randomness will lead RFdiffusion3 to higher-probability structures. Increase this quantity to increase the diversity of designed structures.
-
is_non_loopysetting:The
is_non_loopysetting is a constraint on the designs RFdiffusion3 produces, which makes it a setting provided in a JSON/YAML file. IfTrueit biases the model away from forming structures with many regions without a defined secondary structure. This will slightly decrease the diversity of structures that RFdiffusion3 produces while increasing the designability.
Here are a few plots showing the impacts of these settings in protein-protein interface design tasks:
For the purposes of the plots below:
* `Low temperature` means a `step_scale` of 3 and a `gamma_0` of 0.2.
* Pass rates are refolding pass rates, the number of backbones that pass after four attempts at designing the sequence using MPNN-based methods.
* 'Cluster' refers to `foldseek-based clusters <https://www.nature.com/articles/s41587-023-01773-0>`_, and the cluster pass rate is the number of clusters represented among the passing designs divided by the total number of designed backbones.
:width: 800px
Impacts of using low temperature settings (inf) and the `is_non_loopy` constraint on the outputs of RFdiffusion3.
:width: 800px
Diversity of folds in structures designed by RFD3 when using low temperature sampling and the `is_non_loopy` setting.
:width: 800px
Compares the amount of alpha helices and beta sheets in structures designed by RFD3 when the low temperature sampling and `is_non_loopy` settings are used. The removal of the `is_non_loopy` setting results in a large reduction in α-helices and a small increase in the number of ß-sheets.