actifpTM: A Refined Confidence Metric for AlphaFold2 Predictions Involving Flexible Regions 1. This study introduces actifpTM, a modified interface-predicted TM (ipTM) score, designed to better evaluate the confidence of AlphaFold2 (AF2) predictions, particularly in flexible or disordered regions. 2. Key innovation: actifpTM focuses only on confidently modeled interface residues, addressing biases in traditional ipTM scores that result from flexible flanking regions, making it more robust for peptide-protein interactions. 3. By integrating residue-pair weights based on predicted contact probabilities, actifpTM achieves more accurate confidence evaluations, regardless of flanking region lengths or conformational flexibility. 4. Case studies highlight actifpTM's advantages in assessing peptide-receptor interactions, such as MDM2-p53, SIAH1-CACYBP, and KEAP1-NF2L2, demonstrating more consistent confidence scores compared to ipTM. 5. actifpTM has been implemented into the popular ColabFold pipeline, allowing researchers to leverage this refined metric with simple configuration changes in the Jupyter notebook. 6. This new metric is particularly impactful for peptide-protein modeling, where traditional confidence metrics struggle due to the influence of small binding interfaces and surrounding disordered regions. 7. The study demonstrates how actifpTM can improve the reliability of AF2 predictions, making it a valuable tool for structural biologists and computational biologists working on protein-ligand interactions. @jvarga92 @sokrypton 📜Paper: #AlphaFold2 #ProteinStructure #StructuralBiology #Bioinformatics #MachineLearning #ProteinModeling #PeptideInteractions

שמח לבשר שעדיין ניתן להירשם לכנס הביואינפורמטיקה הישראלי, יום חמישי 27 לאפריל בגבעת רם (למחרת יום הע��מאות) עם שלל דוברות ודוברים מעולים, פאנל על רפואה בעידן הביג דאטה, סדנאות מקצועיות, והרצאות לקהל הרחב ולתלמידי תיכון. בואו בהמוניכם/ן!