LIGYSIS aggregates protein-ligand interactions across biological assemblies of different structures of the same protein trying to get a complete picture of the protein-ligand binding landscape of a given protein!
#LIGYSIS
🔗 : compbio.dundee.ac.uk/ligysis/
📜 : tinyurl.com/utges-lbs
𝗧𝗮𝗸𝗲 𝗵𝗼𝗺𝗲 𝗺𝗲𝘀𝘀𝗮𝗴𝗲𝘀 - 𝗜
1. #LIGYSIS as a new ligand site prediction #reference dataset.
2. #Biological Units > #Asymmetric Units for site prediction test.
3. #Redundancy in prediction #negatively affects performance.
4. Strong #scoring #positively affects performance.
5. #DCC #threshold = 12Å
We compared LIGYSIS to training and test methods used by the methods surveyed in this work: #scPDB, #bindingMOAD, #CHEN11, #PDBbind, #SC6K, #HOLO4K and #COACH420 and #JOINED.
#LIGYSIS (NI) has the highest ligand diversity across datasets.
#PDBbind is diverse and #SC6K is dominated by ATP.
#LIGYSIS avoids this by aggregating #unique biologically #relevant protein-ligand interactions #across #structures, thus representing the most #complete protein-ligand binding dataset to date.
We propose LIGYSIS as a new #benchmark #dataset for #ligand binding #site #prediction.
Our paper "Comparative evaluation of methods for the prediction of protein-ligand binding sites" is now published on Journal of Cheminformatics!
🔗 tinyurl.com/Utges-LBS
On this paper we benchmark 13 original ligand site prediction tools and 15 variants on our curated reference dataset: #LIGYSIS!
Our work should be of interest to those applying fragment screening in drug discovery and more generally in classifying observed and predicted ligand sites from protein structure
The #LIGYSIS dataset builds on this work and was used for our ligand site prediction benchmark!
🔗 tinyurl.com/Utges-LBS
Thought I'd start on @bsky.app with a thread about our paper on characterising #fragment #screening binding sites published on @natureportfolio.bsky.social Comms Bio
This work sets the scene for our current work on #LIGYSIS
Thread is from @gjbarton.bsky.social on X
🔗 tinyurl.com/Utges-FRAGSYS
