Issue 19

We are back just in time for easter! Here is some nice reading for over the (extra long) weekend :).

Errors in structural biology are not the exception

Following the COVID-19 pandemic, structural biologists scrambled to determine the structures of the proteins encoded by SARS-CoV-2 as fast as possbile. Evaluation of these structures by the Coronavirus Structural Task Force revealed that errors in these structures were not the exception. In this scientific comment, Gao et al. discusses errors in structural biology from a broader perspective: Why are errors so common? What are the possible consequences of errors? And what changes do we need to make to the structural biology community to become more resilient to errors?

Protein language models can capture protein quaternary state

Large language models specifically tailored to protein sequences are rapidly becoming more advanced and their usage base is growing. This paper explores if a current protein language model also incorporates information about the arrangement of multiple folded subunits, also known as the quarternary structure, and subsequently builds a model that can classify their number.

AB-Gen: Antibody Library Design with Generative Pre-trained Transformer and Deep Reinforcement Learning

Antibodies can be potent therapeutics for a great number of diseases, which is why a lot of effort is put into designing antibodies that are highly specific and have high affinity for their target. However, the optimization of lead antibodies is a very slow process, which is why in-silico pre-screening of potential leads is of great interest. In this paper, the authors present a method for antibody library design and present a number of test cases.

Some more

• Small molecule machine learning: All models are wrong, some may not even be useful
• Prediction of protein assemblies by structure sampling followed by interface-focused scoring
• Deep Learning for Flexible and Site-Specific Protein Docking and Design

• .@arian_jamasb integrated Foldcomp, our structure compression algorithm, into Graphein - a Geometric Deep Learning framework for protein structures.
  Now, you can train networks on a proteome scale in Colab! Great work. 🎉
  View the notebook: https://t.co/PUPZOgwxc4

  — Martin Steinegger 🇺🇦 (@thesteinegger) March 30, 2023

• Rosetta now also has a diffusion model!
• A useful python package for making deep learning architecture figures.
• Epik: pKa and Protonation State Prediction through Machine Learning
• A general computational design strategy for stabilizing viral class I fusion proteins
• Geometric Deep Learning for Molecular Crystal Structure Prediction

• Hey @Nature, what do you think of this gem from vol. 248, 26 April 1974? What were you all smoking? pic.twitter.com/z7CevnlUSr

  — Matthew Cobb (@matthewcobb) March 19, 2023

• The best of both worlds! Combining AlphaFold2 structures and Rosetta energies allows accurate prediction of thermo(de)stabilization trends when evolving/engineering proteins; very useful for the wet lab! Our last paper in @JCIM_JCTC https://t.co/4Ri02ty3py @CICbioGUNE #compchem

  — Jimenez-Oses Lab (@BilbaoChem) March 18, 2023

• Protein structure prediction with in-cell photo-crosslinking mass spectrometry and deep learning
• VariPred: Enhancing Pathogenicity Prediction of Missense Variants Using Protein Language Models
• ChemEM: flexible docking of small molecules in Cryo-EM structures using difference maps
• Discovering highly potent antimicrobial peptides with deep generative model HydrAMP
• AlphaFlow: autonomous discovery and optimization of multi-step chemistry using a self-driven fluidic lab guided by reinforcement learning