Biology subjects

Langmead, C. J.

Publications and source records attributed to Langmead, C. J..

10 recordsLinked to original sources

bioRxiv2026

pLM representations unlock metagenomic space beyond homology

Metagenomic sequencing has uncovered billions of proteins from uncultured microorganisms, vastly expanding the known protein space. Yet most remain functionally inaccessible because existing annotation methods depend on close homologs or accurate structure predictions. Here, we show that protein language models (pLMs) can unlock this diversity only when their training data are appropriately curated. We introduce Residue Embedding Diversity (RED), a metric for protein quality assessment orders of magnitude cheaper than likelihood, and a calibration task that measures model alignment with natural evolutionary distributions. We discover a fundamental trade-off between evolutionary calibration and structural modeling, establishing training data composition as a primary determinant of pLM behavior. Finally, we successfully retrieve diverse enzyme candidates from billions of metagenomic sequences and validate their expression in vivo.

bioinformatics↗

bioRxiv2026

High-resolution dissection of concept acquisition in different families of protein language models

Protein language models have been increasingly successful on tasks ranging from fitness prediction to functional design, yet what biological knowledge they acquire and where it is encoded within their internal representations remain underexplored. Through a high-resolution layer-by-layer interpretability analysis of 8 models from the ESM2 and AMPLIFY families on 22 concepts from human proteome annotations, we found that these models encode concepts of increasing levels of complexity along their depth: basic physicochemical properties and linear motifs are best captured by early-layer embeddings, secondary structure from subsequent layers, and domain-level semantics from middle layers. Principal component projections of these embeddings showed that they separate biologically meaningful protein groupings, and molecular-biology-inspired interventions demonstrated that pLM embeddings can discriminate phosphomimic-active from inactive mutants. Perhaps surprisingly, we observed that pretraining data and compute had a greater impact on the linear emergence of biological concepts than scaling up parameters. By revealing where biological knowledge is captured in pLMs and which choices shape its emergence, our work offers insights to develop more robust, biologically grounded protein language models.

bioinformatics↗

bioRxiv2025

Mapping the evolution of computationally designed protein binders

Computational protein design enables the generation of binders that target specific epitopes on proteins. However, current approaches often require substantial screening from which hits require further affinity maturation. Methods for experimentally improving designed proteins and exploring their sequence-affinity landscapes could therefore streamline the development of high-affinity binders and inform future design strategies. Here, we use OrthoRep, a system for continuous hypermutation in vivo, to drive the evolution of computationally designed mini protein binders ("minibinders") that target a mammalian receptor. Despite their small sizes (59-72 amino acids), we successfully affinity matured multiple minibinders through strong selection for improved binding and also sampled new regions of minibinder fitness landscapes through extensive neutral drift. One evolved minibinder variant was used to construct a combinatorially complete sequence-affinity map for its six affinity increasing mutations, which revealed nearly full additivity in their contributions to binding. Another minibinder was subjected to both deep mutational scanning and extensive evolution under weak selection, resulting in an evolutionarily diverged collection of binder sequences that revealed non-additive relationships among mutations. Our results highlight that the affinity of computationally designed binders can be rapidly increased through evolution and provide a scalable approach for the evolutionary exploration and subsequent mapping of sequence-affinity landscapes. We suggest that this work will complement protein binder design both as a reliable experimental optimization process and as a vehicle for generating new training data.

synthetic biology↗

bioRxiv2024

Protein Language Models: Is Scaling Necessary?

Public protein sequence databases contain samples from the fitness landscape explored by nature. Protein language models (pLMs) pre-trained on these sequences aim to capture this landscape for tasks like property prediction and protein design. Following the same trend as in natural language processing, pLMs have continuously been scaled up. However, the premise that scale leads to better performance assumes that source databases provide an accurate representation of the underlying fitness landscape, which is likely false. By developing an efficient codebase, designing a modern architecture, and addressing data quality concerns such as sample bias, we introduce AMPLIFY, a best-in-class pLM that is orders of magnitude less expensive to train and deploy than previous models. Furthermore, to support the scientific community and democratize the training of pLMs, we have open-sourced AMPLIFYs pre-training codebase, data, and model checkpoints.

molecular biology↗

bioRxiv2023

Identifying promising sequences for protein engineering using a deep Transformer Protein Language Model

Protein engineers aim to discover and design novel sequences with targeted, desirable properties. Given the near limitless size of the protein sequence landscape, it is no surprise that these desirable sequences are often a relative rarity. This makes identifying such sequences a costly and time-consuming endeavor. In this work, we show how to use a deep Transformer Protein Language Model to identify sequences that have the most promise. Specifically, we use the models self-attention map to calculate a PROMISE SCORE that weights the relative importance of a given sequence according to predicted interactions with a specified binding partner. This PROMISE SCORE can then be used to identify strong binders worthy of further study and experimentation. We use the PROMISE SCORE within two protein engineering contexts-- Nanobody (Nb) discovery and protein optimization. With Nb discovery, we show how the PROMISE SCORE provides an effective way to select lead sequences from Nb repertoires. With protein optimization, we show how to use the PROMISE SCORE to select site-specific mutagenesis experiments that identify a high percentage of improved sequences. In both cases, we also show how the self-attention map used to calculate the PROMISE SCORE can indicate which regions of a protein are involved in intermolecular interactions that drive the targeted property. Finally, we describe how to fine-tune the Transformer Protein Language Model to learn a predictive model for the targeted property, and discuss the capabilities and limitations of fine-tuning with and without knowledge transfer within the context of protein engineering.

bioengineering↗

bioRxiv2022

Gpr88 deletion impacts motivational control independently of striatal dopamine function

BackgroundDisrupted motivational control is a common--but poorly treated--feature of psychiatric disorders. Aberrant mesolimbic dopamine signalling is implicated in motivational symptoms, however direct manipulations to these pathways have yielded suboptimal therapeutic effects. GPR88 is an orphan G protein-coupled receptor highly expressed in the striatum on medium spiny neurons, and therefore well-placed to modulate striatal signalling. While the phenotype of Gpr88 knockout mice supports a disruption of motivational pathways, it is unclear whether GPR88 is involved in reward valuation and/or effort-based decision making in a sex-dependent manner, and if this involves altered dopamine function. MethodsIn male and female Gpr88 knockout mice, we used touchscreen-based progressive ratio, with and without reward devaluation, and effort-related choice tasks to assess motivation and cost/benefit decision making, respectively. To explore whether these motivational behaviours were related to altered striatal dopamine, we quantified expression of dopamine-related genes and/or proteins, and used [18F]DOPA PET and GTP{gamma}[35S] binding to assess pre- and postsynaptic dopamine function, respectively. ResultsWe show that male and female Gpr88 knockout mice display greater motivational drive than wild-type mice, which was maintained following reward devaluation. Further, we show that cost/benefit decision making is impaired in male, but not female, Gpr88 knockout mice. Surprisingly, we found that Gpr88 deletion had no effect on striatal dopamine by any of the measures assessed. ConclusionOur results highlight that GPR88 regulates motivational control of behaviour through a dopaminergic-independent mechanism, providing further support for GPR88 as target for mood symptoms in psychiatric disorders.

neuroscience↗

bioRxiv2020

Differential GLP-1R binding and activation by peptide and non-peptide agonists

Peptide drugs targeting class B1 GPCRs can treat multiple diseases, however there remains substantial interest in the development of orally delivered non-peptide drugs. Here we reveal unexpected overlap between signalling and regulation of the glucagon-like peptide-1 (GLP-1) receptor by the non-peptide agonist, PF 06882961, and GLP-1 that was not observed for another compound, OWL-833. Both compounds are currently in clinical trials for treatment of type 2 diabetes. High resolution cryo-EM structures reveal the binding sites for PF-06882961 and GLP-1 substantially overlap, whereas OWL-833 adopts a unique binding mode with a more open receptor conformation at the extracellular face. Structural differences involving extensive water-mediated hydrogen bond networks could be correlated to functional data to understand how PF 06882961, but not OWL-833, can closely mimic the pharmacological properties of GLP-1. These findings will facilitate rational structure-based discovery of non-peptide agonists targeting class B GPCRs.

pharmacology and toxicology↗

bioRxiv2020

Pharmacological Insights into Safety and Efficacy Determinants for the Development of Adenosine Receptor Biased Agonists in the Treatment of Heart Failure

Adenosine A1 receptors (A1R) are a potential target for cardiac injury treatment due to their cardioprotective/antihypertrophic actions, but drug development has been hampered by on-target side effects such as bradycardia and altered renal haemodynamics. Biased agonism has emerged as an attractive mechanism for A1R-mediated cardioprotection that is haemodynamically safe. Here we investigate the pre-clinical pharmacology, efficacy and side-effect profile of the A1R agonist neladenoson, shown to be safe but ineffective in phase IIb trials for the treatment of heart failure. We compare this agent with the well-characterised, pan-adenosine receptor (AR) agonist NECA, capadenoson, and the A1R biased agonist VCP746, previously shown to be safe and cardioprotective in pre-clinical models of heart failure. We show that like VCP746, neladenoson is biased away from Ca2+ influx relative to NECA and the cAMP pathway at the A1R, a profile predictive of a lack of adenosine-like side effects. Additionally, neladenoson was also biased away from the MAPK pathway at the A1R. In contrast to VCP746, which displays more adenosine-like signalling at the A2BR, neladenoson was a highly selective A1R agonist, with biased, weak agonism at the A2BR. Together these results show that unwanted haemodynamic effects of A1R agonists can be avoided by compounds biased away from Ca2+ influx relative to cAMP, relative to NECA. The failure of neladenoson to reach primary endpoints in clinical trials suggests that A1R-mediated cAMP inhibition may be a poor indicator of effectiveness in chronic heart failure. This study provides additional information that can aid future screening and/or design of improved AR agonists that are safe and efficacious in treating heart failure in patients. ONE-SENTENCE SUMMARYBiased agonists that preference against calcium influx relative to the cyclic AMP pathway, when compared to a conventional agonist, confer clinical safety to A1 adenosine receptor ligands.

pharmacology and toxicology↗

bioRxiv2020

HARVESTMAN: A framework for hierarchical featurelearning and selection from whole genome sequencingdata

We present HO_SCPLOWARVESTMANC_SCPLOW, a method that takes advantage of hierarchical relationships among the possible biological interpretations and representations of genomic variants to perform automatic feature learning, feature selection, and model building. We demonstrate that HO_SCPLOWARVESTMANC_SCPLOW scales to thousands of genomes comprising more than 84 million variants by processing phase 3 data from the 1000 Genomes Project, the largest publicly available collection of whole genome sequences. Next, using breast cancer data from The Cancer Genome Atlas, we show that HO_SCPLOWARVESTMANC_SCPLOW selects a rich combination of representations that are adapted to the learning task, and performs better than a binary representation of SNPs alone. Finally, we compare HO_SCPLOWARVESTMANC_SCPLOW to existing feature selection methods and demonstrate that our method selects smaller and less redundant feature subsets, while maintaining accuracy of the resulting classifier. The data used is available through either the 1000 Genomes Project or The Cancer Genome Atlas. Access to TCGA data requires the completion of a Data Access Request through the Database of Genotypes and Phenotypes (dbGaP). Binary releases of HO_SCPLOWARVESTMANC_SCPLOW compatible with Linux, Windows, and Mac are available for download at https://github.com/cmlh-gp/Harvestman-public/releases

bioinformatics↗

bioRxiv2019

Crystal structure of the M5 muscarinic acetylcholine receptor

The human M5 muscarinic acetylcholine receptor (mAChR) has recently emerged as an exciting therapeutic target for treating a range of disorders, including drug addiction. However, a lack of structural information for this receptor subtype has limited further drug development and validation. Here we report a high-resolution crystal structure of the human M5 mAChR bound to the clinically used inverse agonist, tiotropium. This structure allowed for a comparison across all five mAChR family members that revealed important differences in both orthosteric and allosteric sites that could inform the rational design of selective ligands. These structural studies together with chimeric swaps between the extracellular regions of the M2 and M5 mAChR further revealed the structural insight into \"kinetic-selectivity\", where ligands show differential residency times between related family members. Collectively, our study provides important insights into the nature of orthosteric and allosteric ligand interaction across the mAChR family that could be exploited for the design of selective ligands.\n\nSignificance StatementThe five subtypes of the muscarinic acetylcholine receptors (mAChRs) are expressed throughout the central and peripheral nervous system where they play a vital role in physiology and pathologies. Recently, the M5 mAChR subtype has emerged as an exciting drug target for the treatment of drug addiction. We have determined the atomic structure of the M5 mAChR bound to the clinically used inverse agonist tiotropium. The M5 mAChR structure now allows for a full comparison of all five mAChR subtypes and reveals subtle differences in the extracellular loop (ECL) regions of the receptor that mediate orthosteric and allosteric ligand selectivity. Together these findings open the door for future structure-based design of selective drugs that target this therapeutically important class of receptors.

pharmacology and toxicology↗