Merge remote-tracking branch 'refs/remotes/origin/main'

Author: keiran-rowell-unsw

protein_struct_pred/best-practices-alphafold.md

@@ -1,5 +1,6 @@
 ---
 title: Best practices for presenting and sharing AlphaFold models in a paper
+description: Have you recently started using AlphaFold and want to include its structural predictions in your paper, but don't know the best way to do so? In this short guide, we clarify what you should include so That your work is clear and reproducible for the reader.
 type: guide
 contributors: [James Lingford]
 affiliations: [Monash University]
@@ -20,7 +21,9 @@ For these reasons, we have distilled our knowledge, and that of the broader fiel
 Please reach out to us if you have suggestions.
 
 
-## The minimum AlphaFold information that should be presented in a manuscript
+## What AlphaFold data should I include?
+
+The minimum AlphaFold information that should be presented in a manuscript is the following:
 
 * AlphaFold models should be labelled as such, either in the figure or in the figure legend, so as to avoid being confused with experimentally validated models.
 * The top ranked AlphaFold model should be presented, whereas lower ranked models can be included as supplementary data.
@@ -34,9 +37,9 @@ For example, papers that present cryo-EM structures of a protein will often show
 
 Just like protein models built with cryo-EM, AlphaFold comes with information about model confidence. It is crucial that this information regarding AlphaFold confidence is included somewhere in the paper. The important metrics that should be included are pLDDT scores, PAE plots, pTM scores, and ipTM scores (for protein complexes only). For a short introduction to what these metrics are and how to think about them, please see the EMBL-EBI’s guides to [pLDDT](https://www.ebi.ac.uk/training/online/courses/alphafold/inputs-and-outputs/evaluating-alphafolds-predicted-structures-using-confidence-scores/plddt-understanding-local-confidence/), [PAE](https://www.ebi.ac.uk/training/online/courses/alphafold/inputs-and-outputs/evaluating-alphafolds-predicted-structures-using-confidence-scores/pae-a-measure-of-global-confidence-in-alphafold-predictions/), and [pTM/ipTM](https://www.ebi.ac.uk/training/online/courses/alphafold/inputs-and-outputs/evaluating-alphafolds-predicted-structures-using-confidence-scores/confidence-scores-in-alphafold-multimer/).
 
-## Best practices
+## Best practices for presenting models
 
-### Supplementary figure dedicated to your top ranked AlphaFold model
+### Have a supplementary figure dedicated to your top ranked AlphaFold model
 
 We recommend having a supplementary figure dedicated to showing your top ranked AlphaFold model in cartoon form, with the backbone coloured by pLDDT score confidence, alongside its corresponding PAE plot and pTM/ipTM scores. The colour keys for pLDDT and PAE scores should also be included in this supplemental figure. These key pieces of information are important to communicate the model confidence to the reader, which allows the reader to judge the quality of the model for themselves.
 
@@ -69,7 +72,9 @@ Including a figure of the MSA generated by AlphaFold2 can also communicate valua
 * Healy, M., McNally, K.E., Butkovič, R., Chilton, M. et al. Structure of the endosomal Commander complex linked to Ritscher-Schinzel syndrome (2024) Cell. [https://doi.org/10.1016/j.cell.2023.04.003](https://doi.org/10.1016/j.cell.2023.04.003)
     * **Supp. Fig. 3**
 
-## The minimum AlphaFold methodology info that should be included in the methods section
+## What should I include in the methods section?
+
+The minimum AlphaFold methodology info that should be included in the methods section is the following:
 
 * The specific implementation of AlphaFold used (e.g. AlphaFold2 or 3, ColabFold, LocalColabFold, OpenFold, etc.)
 * All software version numbers 
@@ -80,7 +85,7 @@ Including a figure of the MSA generated by AlphaFold2 can also communicate valua
 
 Like all scientific methodology, the methods section on how AlphaFold was used should be descriptive enough that anyone would be able to replicate your outputs. There are now a few versions of AlphaFold2, as well as “forks” of AlphaFold2 like ColabFold. To ensure reproducibility, it is important to specify exactly what version of the software was used. Likewise, if installation of the software or any dependencies are altered from the original software, it will also be necessary to describe these differences. In such a case, it is probably easier for both the readers and the authors if the authors provide a link to a GitHub repository containing their modified version of the AlphaFold software. This GitHub repository should include a README file with installation and running instructions.
 
-{% include callout.html type="note" content="**Did you know:** using digital object identifiers (DOIs), you can create a link that doesn’t break for software, including modified versions. [See this link to find out how to do this for GitHub](https://docs.github.com/en/repositories/archiving-a-github-repository/referencing-and-citing-content)." %}
+{% include callout.html type="note" content="Did you know that by using digital object identifiers (DOIs), you can create a link that doesn’t break for software, including modified versions. [See this link to find out how to do this for GitHub](https://docs.github.com/en/repositories/archiving-a-github-repository/referencing-and-citing-content)." %}
 
 Describing the parameters used to run AlphaFold2 is very important for proper reproducibility. The AlphaFold2 software has defaults for all their running parameters, but often you will want to change these parameters. Parameters that are often changed include:
 
@@ -100,7 +105,9 @@ Lastly, the creators of any software used should be properly cited. If you are u
 * If you’re running AlphaFold2 via OpenFold, please also cite: Ahdritz, G., Bouatta, N., Floristean, C. et al. OpenFold: retraining AlphaFold2 yields new insights into its learning mechanisms and capacity for generalization. Nat Methods (2024). doi: [10.1038/s41592-024-02272-z](https://www.nature.com/articles/s41592-024-02272-z)
 
 
-## The minimum AlphaFold data to share
+## What AlphaFold data should I share?
+
+The minimum AlphaFold data that we recommend sharing is the following:
 
 * The input amino acid sequence(s) used. Preferably in fasta file format or in a supplementary table. If the input has multiple sequences (i.e. if it’s a multimer), the identities of individual subunits should be clear for the reader.
 * If AlphaFold3 was used, the input .json file should be included. These files can include information about post-translational modifications and ligands that are difficult to communicate in other file formats.