Biology subjects

Steinegger, M.

Publications and source records attributed to Steinegger, M..

At least 55 records · Page 3Linked to original sources

bioRxiv2023

Structural phylogenetics unravels the evolutionary diversification of communication systems in gram-positive bacteria and their viruses

Recent advances in AI-based protein structure modeling have yielded remarkable progress in predicting protein structures. Since structures are constrained by their biological function, their geometry tends to evolve more slowly than the underlying amino acids sequences. This feature of structures could in principle be used to reconstruct phylogenetic trees over longer evolutionary timescales than sequence-based approaches, but until now a reliable structure-based tree building method has been elusive. Here, we introduce a rigorous framework for empirical tree accuracy evaluation and tested multiple approaches using sequence and structure information. The best results were obtained by inferring trees from sequences aligned using a local structural alphabet--an approach robust to conformational changes that confound traditional structural distance measures. We illustrate the power of structure-informed phylogenetics by deciphering the evolutionary diversification of a particularly challenging family: the fast-evolving RRNPPA quorum sensing receptors. We were able to propose a more parsimonious evolutionary history for this critical protein family which enables gram-positive bacteria, plasmids and bacteriophages to communicate and coordinate key behaviors. The advent of high-accuracy structural phylogenetics enables a myriad of applications across biology, such as uncovering deeper evolutionary relationships, elucidating unknown protein functions, or refining the design of bioengineered molecules.

bioinformatics↗

bioRxiv2023

ProstT5: Bilingual Language Model for Protein Sequence and Structure

Adapting large language models (LLMs) to protein sequences spawned the development of powerful protein language models (pLMs). Concurrently, AlphaFold2 broke through in protein structure prediction. Now we can systematically and comprehensively explore the dual nature of proteins that act and exist as three-dimensional (3D) machines and evolve as linear strings of one-dimensional (1D) sequences. Here, we leverage pLMs to simultaneously model both modalities by combining 1D sequences with 3D structure in a single model. We encode protein structures as token sequences using the 3Di-alphabet introduced by the 3D-alignment method Foldseek. This new foundation pLM extracts the features and patterns of the resulting "structure-sequence" representation. Toward this end, we built a non-redundant dataset from AlphaFoldDB and fine-tuned an existing pLM (ProtT5) to translate between 3Di and amino acid sequences. As a proof-of-concept for our novel approach, dubbed Protein structure-sequence T5 (ProstT5), we showed improved performance for subsequent prediction tasks, and for "inverse folding", namely the generation of novel protein sequences adopting a given structural scaffold ("fold"). Our work showcased the potential of pLMs to tap into the information-rich protein structure revolution fueled by AlphaFold2. ProstT5 paves the way to develop new tools integrating the vast resource of 3D predictions, and opens new research avenues in the post-AlphaFold2 era. Our model is freely available for all at https://github.com/mheinzinger/ProstT5.

bioinformatics↗

bioRxiv2023

Petascale Homology Search for Structure Prediction

The recent CASP15 competition highlighted the critical role of multiple sequence alignments (MSAs) in protein structure prediction, as demonstrated by the success of the top AlphaFold2-based prediction methods. To push the boundaries of MSA utilization, we conducted a petabase-scale search of the Sequence Read Archive (SRA), resulting in gigabytes of aligned homologs for CASP15 targets. These were merged with default MSAs produced by ColabFold-search and provided to ColabFold-predict. By using SRA data, we achieved highly accurate predictions (GDT_TS > 70) for 66% of the non-easy targets, whereas using ColabFold-search default MSAs scored highly in only 52%. Next, we tested the effect of deep homology search and ColabFolds advanced features, such as more recycles, on prediction accuracy. While SRA homologs were most significant for improving ColabFolds CASP15 ranking from 11th to 3rd place, other strategies contributed too. We analyze these in the context of existing strategies to improve prediction.

bioinformatics↗

bioRxiv2023

A 2-million-year-old microbial and viral communities from the Kap Kobenhavn Formation in North Greenland

Environmental DNA (eDNA) from the 2-million-year-old Kap Kobenhavn Formation of northern Greenland has revealed an ecosystem of plants and animals with no contemporary analogue1. Here, we reconstruct the microbial (bacterial, archaeal, and viral) communities that thrived at the site during this time. By leveraging a novel analytical framework that integrates taxonomic profiling, DNA damage estimates, and functional reconstructions, we identify and distinguish pioneer microbial communities from later permafrost microbial assemblages. We show that at the time of sediment deposition, the terrestrial input at the Kap Kobenhavn site originated from a palustrine wetland, suggesting warmer, non-permafrost conditions. During this period, the detection of methanogenic archaea and signals of their carbon metabolism is consistent with Kap Kobenhavn and similar northern ecosystems contributing moderate methane emissions. Intriguingly, we discover a remarkable nucleotide sequence similarity--exceeding 98%--between pioneer methanogens and present-day analogues in thawing permafrost. This aligns with the concept of "time-traveling" microbes2 surviving across geological time and waiting for conditions to turn favourable rather than evolving to adapt to changing conditions. Importantly, in contrast to the plant and animal communities of the Kap Kobenhavn, a striking similarity in microbial composition to that of a contemporary thawing Arctic suggests that microbial communities may serve as the first indication of broader climate-driven ecosystem disruptions.

microbiology↗

bioRxiv2023

Metabuli: sensitive and specific metagenomic classification via joint analysis of amino-acid and DNA

Current metagenomic classifiers analyze either DNA or amino-acid (AA) sequences. DNA-based methods have better specificity in distinguishing well-studied clades, but they have limited sensitivity in detecting under-studied clades. AA-based methods suffer the opposite problem. To tackle this trade-off, we developed Metabuli for a joint analysis of DNA and AA using a novel k-mer, metamer. In benchmarks, Metabuli was simultaneously as specific as DNA-based methods and as sensitive as AA-based methods. In the CAMI2 plant-associated dataset, Metabuli covers 99% and 98% of classifications of state-of-the-art DNA-based and AA-based classifiers, respectively. Metabuli is available as free and open-source software for Linux and macOS at metabuli.steineggerlab.com.

bioinformatics↗

bioRxiv2023

Clustering predicted structures at the scale of the known protein universe

Proteins are key to all cellular processes and their structure is important in understanding their function and evolution. Sequence-based predictions of protein structures have increased in accuracy with over 214 million predicted structures available in the AlphaFold database (AFDB). However, studying protein structures at this scale requires highly efficient methods. Here, we developed a structural-alignment based clustering algorithm - Foldseek cluster - that can cluster hundreds of millions of structures. Using this method we have clustered all structures in AFDB, identifying 2.27M non-singleton structural clusters, of which 31% lack annotations representing likely novel structures. Clusters without annotation tend to have few representatives covering only 4% of all proteins in the AFDB. Evolutionary analysis suggests that most clusters are ancient in origin but 4% seem species specific, representing lower quality predictions or examples of de-novo gene birth. Additionally, we show how structural comparisons can be used to predict domain families and their relationships, identifying examples of remote homology. Based on these analyses we identify several examples of human immune related proteins with remote homology in prokaryotic species which illustrates the value of this resource for studying protein function and evolution across the tree of life. AvailabilityMethods and data are available at cluster.foldseek.com

bioinformatics↗

bioRxiv2023

LZTR1 polymerization provokes cardiac pathology in recessive Noonan syndrome

Noonan syndrome patients harboring causative variants in LZTR1 are particularly at risk to develop severe and early-onset hypertrophic cardiomyopathy. However, the underling disease mechanisms of LZTR1 missense variants driving the cardiac pathology are poorly understood. Hence, therapeutic options for Noonan syndrome patients are limited. In this study, we investigated the mechanistic consequences of a novel homozygous causative variant LZTR1L580Pby using patient-specific and CRISPR/Cas9-corrected iPSC-cardiomyocytes. Molecular, cellular, and functional phenotyping in combination with in silico prediction of protein complexes uncovered a unique LZTR1L580P-specific disease mechanism provoking the cardiac hypertrophy. The homozygous variant was predicted to alter the binding affinity of the dimerization domains facilitating the formation of linear LZTR1 polymer chains. The altered polymerization resulted in dysfunction of the LZTR1-cullin 3 ubiquitin ligase complexes and subsequently, in accumulation of RAS GTPases, thereby provoking global pathological changes of the proteomic landscape ultimately leading to cellular hypertrophy. Furthermore, our data showed that cardiomyocyte-specific MRAS degradation is mediated by LZTR1 via the autophagosome, whereas RIT1 degradation is mediated by both LZTR1-dependent and LZTR1-independent proteasomal pathways. Importantly, uni-or biallelic genetic correction of the LZTR1L580P missense variant rescued the molecular and cellular disease-associated phenotype, providing proof-of-concept for CRISPR-based gene therapies.

cell biology↗

bioRxiv2022

CHESS 3: an improved, comprehensive catalog of human genes and transcripts based on large-scale expression data, phylogenetic analysis, and protein structure

The original CHESS database of human genes was assembled from nearly 10,000 RNA sequencing experiments in 53 human body sites produced by the Genotype-Tissue Expression (GTEx) project, and then augmented with genes from other databases to yield a comprehensive collection of protein-coding and noncoding transcripts. The construction of the new CHESS 3 database employed improved transcript assembly algorithms, a new machine learning classifier, and protein structure predictions to identify genes and transcripts likely to be functional and to eliminate those that appeared more likely to represent noise. The new catalog contains 41,356 genes on the GRCh38 reference human genome, of which 19,839 are protein-coding, and a total of 158,377 transcripts. These include 14,863 novel protein-coding transcripts. The total number of transcripts is substantially smaller than earlier versions due to improved transcriptome assembly methods and to a stricter protocol for filtering out noisy transcripts. Notably, CHESS 3 contains all of the transcripts in the MANE database, and at least one transcript corresponding to the vast majority of protein-coding genes in the RefSeq and GENCODE databases. CHESS 3 has also been mapped onto the complete CHM13 human genome, which gives a more-complete gene count of 43,773 genes and 19,968 protein-coding genes. The CHESS database is available at http://ccb.jhu.edu/chess.

genomics↗

bioRxiv2022

Foldcomp: a library and format for compressing and indexing large protein structure sets

Highly accurate protein structure predictors have generated hundreds of millions of protein structures; these pose a challenge in terms of storage and processing. Here we present Foldcomp, a novel lossy structure compression algorithm and indexing system to address this challenge. By using a combination of internal and cartesian coordinates and a bi-directional NeRF-based strategy, Foldcomp improves the compression ratio by a factor of 3 compared to the next best method. Its reconstruction error of 0.08[A] is comparable to the best lossy compressor. It is 5 times faster than the next fastest compressor and competes with the fastest decompressors. With its multi-threading implementation and a Python interface that allows for easy database downloads and efficient querying of protein structures by accession, Foldcomp is a powerful tool for managing and analyzing large collections of protein structures. AvailabilityFoldcomp is a free open-source library and command-line software available for Linux, macOS and Windows at https://foldcomp.foldseek.com. Foldcomp provides the AlphaFold Swiss-Prot (2.9GB), TrEMBL (1.1TB) and ESMatlas HQ (114GB) database ready-for-download.

bioinformatics↗

bioRxiv2022

Ultra-fast protein structure prediction to capture effects of sequence variation in mutation movies

Top protein three-dimensional (3D) structure predictions require evolutionary information from multiple-sequence alignments (MSAs) and deep, convolutional neural networks and appear insensitive to small sequence changes. Here, we describe EMBER3D using embeddings from the pre-trained protein language model (pLM) ProtT5 to predict 3D structure directly from single sequences. Orders of magnitude faster than others, EMBER3D predicts average-length structures in milliseconds on consumer-grade machines. Although not nearly as accurate as AlphaFold2, the speed of EMBER3D allows a glimpse at future applications such as the almost real-time rendering of deep mutational scanning (DMS) movies that visualize the effect of all point mutants on predicted structures. This also enables live-editing of sequence/structure pairs. EMBER3D is accurate enough for highly sensitive rapid remote homology detection by Foldseek identifying structural similarities. Overall, our use cases suggest that speed can complement accuracy, in particular when accessible through consumer-grade machines. EMBER3D is free and publicly available: https://github.com/kWeissenow/EMBER3D.

bioinformatics↗

bioRxiv2022

Nearest neighbor search on embeddings rapidly identifies distant protein relations

Since 1992, all state-of-the-art (SOTA) methods for fast and sensitive identification of evolutionary, structural, and functional relations between proteins (also referred to as "homology detection") use sequences and sequence-profiles (PSSMs). Protein Language Models (pLMs) generalize sequences, possibly capturing the same constraints as PSSMs, e.g., through embeddings. Here, we explored how to use such embeddings for nearest neighbor searches to identify relations between protein pairs with diverged sequences (remote homology detection for levels of <20% pairwise sequence identity, PIDE). While this approach excelled for proteins with single domains, we demonstrated the current challenges applying this to multi-domain proteins and presented some ideas how to overcome existing limitations, in principle. We observed that sufficiently challenging data set separations were crucial to provide deeply relevant insights into the behavior of nearest neighbor search when applied to the protein embedding space, and made all our methods readily available for others.

bioinformatics↗

bioRxiv2022

UFCG: database of universal fungal core genes and pipeline for genome-wide phylogenetic analysis of fungi

In phylogenomics the evolutionary relationship of organisms is studied by their genomic information. A common approach to phylogenomics is to extract related genes from each organism, build a multiple sequence alignment and then reconstruct evolution relations through a phylogenetic tree. Often a set of highly conserved genes occurring in single-copy, called core genes, are used for this analysis, as they allow efficient automation within a taxonomic clade. Here we introduce the Universal Fungal Core Genes (UFCG) database and pipeline for genome-wide phylogenetic analysis of fungi. The UFCG database consists of 61 curated fungal marker genes, including a novel set of 41 computationally derived core genes and 20 canonical genes derived from literature, as well as marker gene sequences extracted from publicly available fungal genomes. Furthermore, we provide an easy-to-use, fully automated and open-source pipeline for marker gene extraction, training and phylogenetic tree reconstruction. The UFCG pipeline can identify marker genes from genomic, proteomic and transcriptomic data, while producing phylogenies consistent with those previously reported, and is publicly available together with the UFCG database at https://ufcg.steineggerlab.com.

bioinformatics↗

bioRxiv2022

LambdaPP: Fast and accessible protein-specific phenotype predictions

The availability of accurate and fast Artificial Intelligence (AI) solutions predicting aspects of proteins are revolutionizing experimental and computational molecular biology. The webserver LambdaPP aspires to supersede PredictProtein, the first internet server making AI protein predictions available in 1992. Given a protein sequence as input, LambdaPP provides easily accessible visualizations of protein 3D structure, along with predictions at the protein level (GeneOntology, subcellular location), and the residue level (binding to metal ions, small molecules, and nucleotides; conservation; intrinsic disorder; secondary structure; alpha-helical and beta-barrel transmembrane segments; signal-peptides; variant effect) in seconds. The structure prediction provided by LambdaPP - leveraging ColabFold and computed in minutes - is based on MMseqs2 multiple sequence alignments. All other feature prediction methods are based on the pLM ProtT5. Queried by a protein sequence, LambdaPP computes protein and residue predictions almost instantly for various phenotypes, including 3D structure and aspects of protein function. Accessibility StatementLambdaPP is freely available for everyone to use under embed.predictprotein.org, the interactive results for the case study can be found under https://embed.predictprotein.org/o/Q9NZC2. The frontend of LambdaPP can be found on GitHub (github.com/sacdallago/embed.predictprotein.org), and can be freely used and distributed under the academic free use license (AFL-2). For high-throughput applications, all methods can be executed locally via the bio-embeddings (bioembeddings.com) python package, or docker image at ghcr.io/bioembeddings/bio_embeddings, which also includes the backend of LambdaPP. Impact StatementWe introduce LambdaPP, a webserver integrating fast and accurate sequence-only protein feature predictions based on embeddings from protein Language Models (pLMs) available in seconds along with high-quality protein structure predictions. The intuitive interface invites experts and novices to benefit from the latest machine learning tools. LambdaPPs unique combination of predicted features may help in formulating hypotheses for experiments and as input to bioinformatics pipelines.

bioinformatics↗

bioRxiv2022

3D-Beacons: Decreasing the gap between protein sequences and structures through a federated network of protein structure data resources

While scientists can often infer the biological function of proteins from their 3-dimensional quaternary structures, the gap between the number of known protein sequences and their experimentally determined structures keeps increasing. A potential solution to this problem is presented by ever more sophisticated computational protein modelling approaches. While often powerful on their own, most methods have strengths and weaknesses. Therefore, it benefits researchers to examine models from various model providers and perform comparative analysis to identify what models can best address their specific use cases. To make data from a large array of model providers more easily accessible to the broader scientific community, we established 3D-Beacons, a collaborative initiative to create a federated network with unified data access mechanisms. The 3D-Beacons Network allows researchers to collate coordinate files and metadata for experimentally determined and theoretical protein models from state-of-the-art and specialist model providers and also from the Protein Data Bank.

bioinformatics↗

bioRxiv2022

Highly accurate isoform identification for the human transcriptome

We explore a new hypothesis in genome annotation, namely whether computationally predicted protein structures can help to identify which of multiple possible gene isoforms represents a functional protein product. Guided by structure predictions, we evaluated over 140,000 isoforms of human protein-coding genes assembled from over 10,000 RNA sequencing experiments across many human tissues. We illustrate our new method with examples where structure provides a guide to function in combination with expression and evolutionary evidence. Additionally, we provide the complete set of structures as a resource to better understand the function of human genes and their isoforms. These results demonstrate the promise of protein structure prediction as a genome annotation tool, allowing us to refine even the most highly-curated catalog of human proteins. One-Sentence SummaryWe describe the use of 3D protein structures on a genome-wide scale to evaluate human protein isoforms for biological functionality.

genomics↗

bioRxiv2022

AlphaFold2 reveals commonalities and novelties in protein structure space for 21 model organisms

Over the last year, there have been substantial improvements in protein structure prediction, particularly in methods like DeepMinds AlphaFold2 (AF2) that exploit deep learning strategies. Here we report a new CATH-Assign protocol which is used to analyse the first tranche of AF2 models predicted for 21 model organisms and discuss insights these models bring on the nature of protein structure space. We analyse good quality models and those with no unusual structural characteristics, i.e., features rarely seen in experimental structures. For the [~]370,000 models that meet these criteria, we observe that 92% can be assigned to evolutionary superfamilies in CATH. The remaining domains cluster into 2,367 putative novel superfamilies. Detailed manual analysis on a subset of 618 of those which had at least one human relative revealed some extremely remote homologies and some further unusual features, but 26 could be confirmed as novel superfamilies and one of these has an alpha-beta propeller architectural arrangement never seen before. By clustering both experimental and predicted AF2 domain structures into distinct global fold groups, we observe that the new AF2 models in CATH increase information on structural diversity by 36%. This expansion in structural diversity will help to reveal associated functional diversity not previously detected. Our novel CATH-Assign protocol scales well and will be able to harness the huge expansion (at least 100 million models) in structural data promised by DeepMind to provide more comprehensive coverage of even the most diverse superfamilies to help rationalise evolutionary changes in their functions.

bioinformatics↗

bioRxiv2022

Foldseek: fast and accurate protein structure search

As structure prediction methods are generating millions of publicly available protein structures, searching these databases is becoming a bottleneck. Foldseek aligns the structure of a query protein against a database by describing the amino acid backbone of proteins as sequences over a structural alphabet. Foldseek decreases computation times by four to five orders of magnitude with 86%, 88% and 133% of the sensitivities of DALI, TM-align and CE, respectively.

bioinformatics↗

bioRxiv2021

Block aligner: fast and flexible pairwise sequence alignment with SIMD-accelerated adaptive blocks

BackgroundThe Smith-Waterman-Gotoh alignment algorithm is the most popular method for comparing biological sequences. Recently, Single Instruction Multiple Data methods have been used to speed up alignment. However, these algorithms have limitations like being optimized for specific scoring schemes, cannot handle large gaps, or require quadratic time computation. ResultsWe propose a new algorithm called block aligner for aligning nucleotide and protein sequences. It greedily shifts and grows a block of computed scores to span large gaps within the aligned sequences. This greedy approach is able to only compute a fraction of the DP matrix. In exchange for these features, there is no guarantee that the computed scores are accurate compared to full DP. However, in our experiments, we show that block aligner performs accurately on various realistic datasets, and it is up to 9 times faster than the popular Farrars algorithm for protein global alignments. ConclusionsOur algorithm has applications in computing global alignments and X-drop alignments on proteins and long reads. It is available as a Rust library at https://github.com/Daniel-Liu-c0deb0t/block-aligner.

bioinformatics↗