These are many .wts file in database\scoring\weights.
I don't clearly know which weight is used in which occasions,
and just use score function according to papers.
I read the webpage(https://www.rosettacommons.org/demos/latest/tutorials/scoring/scoring)
and its recommend references.
It seem that talaris2013，talaris2014, ref2015 and so on are "standard scorefunction".
But those papers seemingly do not describe the development of ligand score function.
（such as ligand， ligand_soft_rep, ligandprime and more）
I just want to kown the differences of weights, which weight is newer and older, and which weight should be use in a specific occasion.
Could answer my question or give some recommend references？
The ligand score function was developed for ligand docking by Meiler and Davis, as outlined in the original RosettaLigand papers: (https://www.ncbi.nlm.nih.gov/pubmed/16972285 and https://www.ncbi.nlm.nih.gov/pubmed/19041878 primariliy). These were developed in the pre-talaris ("score12") environment, so to be strictly correct, you'll want to enable the -restore_pre_talaris_2013_behavior option when using these scorefunctions.
The difference between ligand.wts and ligand_sot_rep.wts is the difference between hardrep and softrep. The hardrep scorefunction is the standard scorefunction, and what you'll want to use whenever you're minimizing. The softrep version of a scorefunction has an altered repulsive potential that's supposed to be more forgiving (softer). The reason for this is that when you're doing discrete rotamer repacking, you can often end up with very small clashes which can be easily resolved by slight minimization. However, the discrete rotamer repacking doesn't allow you to relieve those clashes, so you end up unfairly discarding those rotamers. The softrep function allows a little more "give", permitting you to pack or design in sidechains which are bad "on-rotamer" but can be easily fixed by minimizing slighlty off rotamer.
The ligandprime.wts is a hard rep scorfunction equivalent to ligand.wts, but has the amino acid reference energies refitted using the same proceedure used for reference energy refitting for talaris2013. (See https://www.ncbi.nlm.nih.gov/pubmed/23422428 for details.)
When should you use each? Currently (Jun 2018), the ligand specific weights (with -restore_pre_talaris_2013_behavior enabled) is the recommended scorefunction to use with ligand docking - that is, when attempting to find the binding position of a small molecule in a protein pocket. See https://www.ncbi.nlm.nih.gov/pubmed/26207742 and Rosetta/demos/public/rosettaligand_transform/ for the currently recommended docking proceedure. In this protocol, the ligand_soft_rep.wts (slightly modified) is used during repacking, and the ligand.wts file (slightly modlfied) is used for minimization.
The ligandprime.wts would only be useful (only make a difference) for design. However, current tests indicate that for design of the protein the standard protein scorefunction (ref2015) does as good/better than the ligand-specific scorefunction. (It's all about which side you're samplling. When sampling the ligand side, the ligand-specific scorefunction is better. When sampling the protein side, the most recent protein-specific scorefunction is better.) I wouldn't necessarily recommend using the ligandprime.wts scorefunction when designing the protein side.
I will note that there's currently an effort underway to re-evaluate and update the current ligand docking scorefunctions. There's not yet much to divulge, but hopefully within a year we'll have an update scorefunction for ligand docking which combines the best of the updated protein scorefunctions with the performance on ligand docking of the ligand scorefunction.
Thank you for your detailed reply~
I learn more from your reply and will read those paper you provide.
Looking forward to the update scorefunction ~