Claus O. Wilke

Inherent limitations of zero-shot mutation prediction in protein engineering

This study evaluates zero-shot mutation prediction models in protein engineering, finding they struggle to identify optimal mutations and predict new functions due to limitations in capturing evolutionary constraints.

A structure-guided classification framework reveals the diversity and catalytic architecture of BECR ribonuclease

The paper classifies BECR ribonuclease proteins across domains of life using structural and genomic analysis, revealing new families and refining existing classifications.

Scaling SMILES-Based Chemical Language Models for Therapeutic Peptide Engineering

This paper introduces PeptideCLM-2, a chemical language model for therapeutic peptides, showing strong performance in predicting drug development endpoints.

Intrinsic dataset features drive mutational effect prediction by protein language models

The paper evaluates protein language models' performance on mutational effect prediction across datasets and finds that dataset features, not model architecture, primarily drive predictive success.

A Linear Mixed Effects Model for Evaluating Synthetic Gene Circuits

The paper introduces a linear mixed effects model to statistically evaluate synthetic gene circuits, finding that the fixed effect β describes gate behavior and can be used to assess performance.

Availability of charged tRNAs drives maximal protein synthesis at intermediate levels of codon usage bias

The paper develops a mathematical model of translation incorporating tRNA dynamics and re-charging to show that codon usage optimizes protein synthesis depending on tRNA availability.

A fold switch regulates conformation of an alphavirus virus RNA-dependent RNA polymerase

This study identifies a fold switch in ONNV nsP4 that regulates its conformation, RNA binding, and polymerase activity through proteolytic cleavage and fold switching.

Medium-sized protein language models perform well at transfer learning on realistic datasets

Medium-sized protein language models (e.g. ESM C 600M) with mean embeddings nearly match ESM-2 15B and ESM C 6B in transfer learning on mutational scanning and PISCES data.

A systematic evaluation of the language-of-viral-escape model using multiple machine learning frameworks

The paper evaluates a language-of-viral-escape model using multiple machine learning frameworks to identify immune escape mutations in viruses.

Peptide-aware chemical language model successfully predicts membrane diffusion of cyclic peptides

The paper introduces PeptideCLM, a chemical language model that predicts membrane diffusion of cyclic peptides more effectively than existing models.

Modeling and measuring how codon usage modulates the relationship between burden and yield during protein overexpression in bacteria

The paper develops a stochastic gene expression model to quantify how codon usage bias affects burden and yield during protein overexpression in bacteria, finding an overoptimization domain that reduces yield.

A computational model for bacteriophage {varphi}X174 gene expression

This study constructs a computational model of bacteriophage φX174 gene regulation and shows that canonical promoters and terminators suffice to explain transcript abundance variation.

Single-cell virology: On-chip, quantitative characterization of the dynamics of virus spread from one single cell to another

This study uses a microfluidic device to characterize lytic and non-lytic virus spread between single cells, finding non-lytic spread is faster and more efficient.

Distinct horizontal transfer mechanisms for type I and type V CRISPR-associated transposons

This study identifies distinct horizontal transfer mechanisms for type I and type V CRISPR-associated transposons, showing they use CRISPR RNAs and alternative unguided mechanisms for gene transfer.

Metagenomic Discovery of CRISPR-Associated Transposons

The paper identifies new CRISPR-associated transposons (CASTs) by analyzing metagenomic databases, revealing novel architectures and systems that co-opt different CRISPR-Cas effectors.

Learning the local landscape of protein structures with convolutional neural networks

This study uses 3D CNNs to predict wild-type amino acids and consensus sequences from local structural context, finding high-confidence predictions correlate with correct outcomes.

BACPHLIP: Predicting bacteriophage lifestyle from conserved protein domains

BACPHLIP predicts bacteriophage lifestyle using conserved protein domains with 98% accuracy, outperforming existing tools.

Virulent but not temperate bacteriophages display hallmarks of rapid translation initiation

The paper shows virulent bacteriophages have mRNAs with strong Shine-Dalgarno sequences and weak start codon structures, suggesting rapid translation, unlike temperate phages.

A systematic analysis of the beta hairpin motif in the Protein Data Bank

The paper systematically analyzes 49,000 beta hairpin substructures from the PDB to identify characteristics correlated with stability and provides design principles for synthetic protein libraries.

Generating complex patterns of gene expression without regulatory circuits

The paper demonstrates that complex gene expression patterns can be achieved in bacteriophage genomes by adjusting promoter, terminator, and RNase cleavage site placements and strengths without dedicated regulatory genes.

Uncovering modeling features of viral replication dynamics from high-throughput single-cell virology experiments

This study develops a stochastic mechanistic model to analyze viral replication dynamics from single-cell poliovirus infection experiments, identifying early replication steps as key drivers of variability in viral growth.

Humanization of yeast genes with multiple human orthologs reveals principles of functional divergence between paralogs

The paper replaces yeast genes with human orthologs to study functional divergence, finding that replaceability depends on divergence and expression patterns of human co-orthologs.

Transcript degradation and codon usage regulate gene expression in a lytic phage

This study shows that transcript degradation and codon usage regulate gene expression in bacteriophage T7 through a dynamic interplay between transcription, translation, and transcript degradation.

Pinetree: a step-wise gene expression simulator with codon-specific translation rates

Pinetree simulates gene expression with codon-specific translation rates and provides a Python interface for genome-based simulations.

Predicting bacterial growth conditions from mRNA and protein abundances

This study uses machine learning to predict bacterial growth conditions from mRNA and protein abundances in E. coli, showing that combined transcriptome and proteome data improve accuracy.

Theory of measurement for site-specific evolutionary rates in amino-acid sequences

The paper develops a theory for measuring site-specific evolutionary rates in amino-acid sequences using maximum likelihood under a mutation-selection model, showing that rates measured relative to a naive exchangeability matrix recover true expected rates.

Limitation of alignment-free tools in total RNA-seq quantification

This study compares alignment-free and alignment-based RNA-seq pipelines for total RNA quantification, finding alignment-free tools less accurate for small and lowly-expressed RNAs.

Reduced protein expression in a virus attenuated by codon deoptimization

The paper investigates how codon deoptimization in bacteriophage T7 reduces protein expression and viral fitness by halving translation of key genes.

The E. coli molecular phenotype under different growth conditions

This study presents a comprehensive transcriptomics and proteomics dataset of E. coli under 34 growth conditions, revealing differences between exponential and stationary phases and differential gene expression under specific stresses.

Computational prediction of the tolerance to amino-acid deletion in green-fluorescent protein

The paper predicts eGFP's tolerance to amino-acid deletions using structural properties, finding that weighted contact number is the best single predictor.