Guillermo Pérez Hernández (@Guille_PH) header
Guillermo Pérez Hernández (@Guille_PH) profile picture

https://nitter.cf/t.co/ZsjIooDLYo https://nitter.cf/t.co/9F25vwL9e6 https://nitter.cf/t.co/zwCLSZxmA8

Berlin
Joined April 2011
Photo by Guillermo Pérez Hernández (@Guille_PH)Photo by Guillermo Pérez Hernández (@Guille_PH)Photo by Guillermo Pérez Hernández (@Guille_PH)Photo by Guillermo Pérez Hernández (@Guille_PH)
Our latest article is out 🎉 @HosseinBatebi & Peter Hildebrand @UniLeipzig! nature.com/articles/s41594-0… This @NatureSMB complements our previous @Nature article led by @MPapasergiScott & #LabSkiniotis I’m opening my first thread ever for some highlights! #GPCR #MD 🧵 👇 [1/n]
3
12
2
23
2,617
Guillermo Pérez Hernández retweeted
Some more evidence that BioEmu et al have largely rendered clever AF2 template and MSA hacks obsolete for quick conformational sampling of soluble monomeric proteins
11
1
66
6,179
Guillermo Pérez Hernández retweeted
🚀 ¡Éxito total! Gracias a tod@s los que nos acompañaron en la visita a BESSY II y SeaLab. Ciencia, curiosidad y aprendizaje de la mano de los Drs. Pablo Echeverría y Adolfo Vélez. #Ciencia #CERFA
2
1
238
Happy to see our ipSAE score for AF3 models is useful for de novo binder design.
Predicting Experimental Success in De Novo Binder Design: A Meta-Analysis of 3,766 Experimentally Characterised Binders 1. A new large-scale meta-analysis of 3,766 experimentally tested de novo protein binders targeting 15 diverse proteins reveals key insights into predicting binding success. The study shows that interface-focused metrics, especially the AF3-derived interaction prediction Score from Aligned Errors (ipSAE), significantly outperform commonly used scores like ipAE and ipTM, achieving a 1.4-fold increase in average precision. 2. The research highlights that combining ipSAE with orthogonal physicochemical interface descriptors, such as Rosetta ∆G/∆SASA and interface shape complementarity, further improves predictive performance. This combination strategy enhances the precision of selecting binders for experimental testing, offering a more reliable approach than traditional methods. 3. The study developed a streamlined pipeline using AF2, AF3, and Boltz-1 to efficiently re-predict binder–target complexes and extract over 200 structural, energetic, and confidence features per design. This pipeline enables high-throughput evaluation with minimal loss in model confidence, making it practical for large-scale benchmarking. 4. The analysis demonstrates significant variability in predictive performance across different targets, with precision ranging from 0.1 to 1.0. However, the study identifies that targets with more energetically intense binding interfaces (e.g., higher H-bond percentage or more negative ∆G/∆SASA) tend to be more predictable, providing potential guidelines for future binder design campaigns. 5. The researchers propose interpretable, target-agnostic filtering strategies, such as combining AF3 ipSAE_min rankings with structural filters, to improve precision in selecting binders. These strategies are supported by a comprehensive dataset released as a community resource to benchmark and accelerate de novo binder discovery. 📜Paper: biorxiv.org/content/10.1101/… #DeNovoBinderDesign #ProteinEngineering #MetaAnalysis #Bioinformatics #PredictiveModeling
2
7
81
5,689
Guillermo Pérez Hernández retweeted
Please RT! I have 2 postdoctoral positions in my lab @BRFAA_IIBEAA to work on MD sims for protein allostery, cryptic pocket identification, CADD/FEP. Knowledge in coarse-grained MD is a plus. More info: linkedin.com/hiring/jobs/424…. Send me an email with your CV if interested!
1
70
2
118
12,266
Guillermo Pérez Hernández retweeted
Desde CERFA, un año más, os deseamos una Feliz Navidad y un próspero año Nuevo. Que todos vuestros deseos científicos para el 2025 se hagan realidad. Gracias por este maravilloso año. Equipo CERFA
3
8
267
Guillermo Pérez Hernández retweeted
Join us next Friday November 29th at the Instituto Cervantes in Berlin! Learn what happens inside your brain when you start speaking, when you learn a new language and why we lose the capacity to speak a language.
3
4
287
Guillermo Pérez Hernández retweeted
Our recent work on microalgae based materials for 3D printing has been highlighted in National Geographic Germany!!! 😍 @claravamar_ @Cluster3DMM2O 🌱
1
3
57
2,749
Guillermo Pérez Hernández retweeted
Hoy estuvo con nosotros Alberto Hernández, catedrático del departamento de Química de la @ULL @CienciaULL
2
8
1
18
740
Guillermo Pérez Hernández retweeted
ShEPhERD: Diffusing shape, electrostatics, and pharmacophores for bioisosteric drug design #compchem shorturl.at/v4QM1
9
22
1,626
Making Berlin a cooler place to do science
immensely grateful to the @Tagesspiegel Top 100 Minds in Science for this honor to the Kraushar Lab @MPI_MolGen, my mentors, colleagues, scientific community, friends and family near and far, and Berlin - thank you for becoming home ❤️
1
3
165
Guillermo Pérez Hernández retweeted
Finalizamos la sesión I: AI & Environment con Mario Mesas Rodriguez que es AI Engineer en BMW @BMW
2
6
214
Ya tenemos ganadora de los premios @acesuecia ACES Margarita Salas! Que ilusión anunciarlos como cada anyo y ya van 4! Muchas gracias a @MsalasFundacion @FundacionAreces @EmbEspEstocolmo por su apoyo continuo, y enhorabuena a @SociedadCerfa y a @raicex_es por las candidaturas
La Profesora Beatriz Roldán Cuenya, Directora del Departamento de Ciencia Interfacial del Instituto Fritz Haber @maxplanckpress en Berlín (Alemania) ganadora de la IV Edición del Premio “ACES – Margarita Salas” Física, Matemáticas e Ingeniería aces-sffs.com/premios-aces-m…
2
5
24
1,726
Guillermo Pérez Hernández retweeted
La Profesora Beatriz Roldán Cuenya, Directora del Departamento de Ciencia Interfacial del Instituto Fritz Haber @maxplanckpress en Berlín (Alemania) ganadora de la IV Edición del Premio “ACES – Margarita Salas” Física, Matemáticas e Ingeniería aces-sffs.com/premios-aces-m…
1
10
1
22
2,704
Guillermo Pérez Hernández retweeted
El simposio ya está aquí! Durante esta semana iremos introduciendo a los ponentes. Empezamos con la sesión I: AI & Environment y la Dra. Mónica González. Postdoc en el Environment group en el Center of Allergy and Environment (ZAUM, @Allergy_ZAUM) TUM-Helmholtz Center, Munich
1
4
119
Guillermo Pérez Hernández retweeted
No os perdáis las charlas de nuestros dos CERFERS (Verónica Díez @Daniajinn y Guillermo Pérez @Guille_PH ) en la Berlin Science Week a las 15:00 en el evento en inglés organizado por la Embajada de España sobre "Digitising Dinosaurs". Más info aquí: buff.ly/40wyqk0
1
2
6
207
Guillermo Pérez Hernández retweeted
Y por fin tenemos el cartel definitivo para nuestro Simposio de este año. En lo próximos días os iremos introduciendo a los ponentes uno a uno. Si aún no te has registrado puedes hacerlo aquí: buff.ly/4diEO0I Te esperamos!
5
1
9
935
Guillermo Pérez Hernández retweeted
Excited to share my and @sonyahans's little guide to large parallel simulations (and what you can do with them) in preprint form!
Running and Analyzing Massively Parallel Molecular Simulations 1. This article provides a comprehensive guide on generating and analyzing massively parallel molecular dynamics (MD) simulations to explore protein conformational landscapes and their role in biological processes. 2. Massively parallel simulations overcome the limitations of single MD simulations by sampling long-timescale events and complex biological phenomena through distributed computing. 3. The paper highlights the use of platforms like Folding@home to create datasets from thousands of citizen-scientist-powered simulations, capturing entire energy landscapes of proteins at unprecedented scales. 4. It emphasizes key practical considerations, including selecting appropriate force fields, handling storage needs, and configuring simulations to maintain consistency across datasets for future research comparisons. 5. The guide introduces advanced methods such as adaptive sampling and replica exchange, which improve the exploration of rare events and enhance predictive power for therapeutic design. 6. The analysis section details how Markov state models (MSMs) and time-lagged independent component analysis (tICA) enable insights into protein dynamics, allostery, and mutational impacts relevant to health and disease. 7. Tools like OpenMM and GROMACS are recommended for running large-scale MD simulations, with detailed advice on handling complex systems such as membrane proteins or protein-ligand complexes. 8. The article also explores novel strategies for handling computational challenges, such as varying initial configurations and using adaptive seeding to ensure robust sampling. 💻Code: github.com/sukritsingh/massi… 📜Paper: doi.org/10.26434/chemrxiv-20…
13
34
2,892