The DOI matches its Crossref record; 31 of the 31 references it cites were found in a registry or an index.
Protein-level assembly increases protein sequence recovery from metagenomic samples manyfold
Open accessPublished version of a preprint
Abstract
The open-source de novo protein-level assembler, Plass ( https://plass.mmseqs.com ), assembles six-frame-translated sequencing reads into protein sequences. It recovers 2–10 times more protein sequences from complex metagenomes and can assemble huge datasets. We assembled two redundancy-filtered reference protein catalogs, 2 billion sequences from 640 soil samples (soil reference protein catalog) and 292 million sequences from 775 marine eukaryotic metatranscriptomes (marine eukaryotic reference catalog), the largest free collections of protein sequences. The protein-level assembler can assemble protein catalogs from raw metagenomic sequencing data, enabling large-scale metagenomics studies.
Overview
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
FLASH: fast length adjustment of short reads to improve genome assemblies
Magoč, T., & Salzberg, S. L. (2011) · Bioinformatics · doi:10.1093/bioinformatics/btr507 · Cited by 15,366
Prodigal: prokaryotic gene recognition and translation initiation site identification
Hyatt, D., Chen, G.-L., LoCascio, P. F., Land, M., Larimer, F. W., & Hauser, L. (2010) · BMC Bioinformatics · doi:10.1186/1471-2105-11-119 · Cited by 12,692
Li, D., Liu, C.-M., Luo, R., Sadakane, K., & Lam, T. (2015) · Bioinformatics · doi:10.1093/bioinformatics/btv033 · Cited by 10,133
Velvet: Algorithms for de novo short read assembly using de Bruijn graphsPreprint
Zerbino, D. R., & Birney, E. (2008) · Genome Research · doi:10.1101/gr.074492.107 · Cited by 9,587
KEGG: new perspectives on genomes, pathways, diseases and drugs
Kanehisa, M., Furumichi, M., Tanabe, M., Sato, Y., & Morishima, K. (2017) · Nucleic Acids Research · doi:10.1093/nar/gkw1092 · Cited by 8,865
MMseqs2 enables sensitive protein sequence searching for the analysis of massive data sets
Steinegger, M., & Söding, J. (2017) · Nature Biotechnology · doi:10.1038/nbt.3988 · Cited by 4,840
Self-citationmetaSPAdes: a new versatile metagenomic assembler
Nurk, S., Meleshko, D., Korobeynikov, A., & Pevzner, P. A. (2017) · Genome Research · doi:10.1101/gr.213959.116 · Cited by 4,496
Structure and function of the global ocean microbiome
Sunagawa, S., Coelho, L. P., Chaffron, S., Kultima, J. R., Labadie, K., Salazar, G., Djahanschiri, B., Zeller, G., Mende, D. R., Alberti, A., et al. (2015) · Science · doi:10.1126/science.1261359 · Cited by 2,978
Fast Genome-Wide Functional Annotation through Orthology Assignment by eggNOG-Mapper
Huerta-Cepas, J., Forslund, K., Coelho, L. P., Szklarczyk, D., Jensen, L. J., von Mering, C., & Bork, P. (2017) · Molecular Biology and Evolution · doi:10.1093/molbev/msx148 · Cited by 2,709
An integrated catalog of reference genes in the human gut microbiome
MetaHIT Consortium, Li, J., Jia, H., Cai, X., Zhong, H., Feng, Q., Sunagawa, S., Arumugam, M., Kultima, J. R., Prifti, E., et al. (2014) · Nature Biotechnology · doi:10.1038/nbt.2942 · Cited by 2,164
Cited by
Sorted by citations
Items 101–200 of 446
Recent advances in structure‐based enzyme engineering for functional reconstruction
Xu, S., Zhou, L., Xu, Y., Hong, H., Dai, C., Wang, Y., & Zheng, Y. (2023) · Biotechnology and Bioengineering · doi:10.1002/bit.28540 · Cited by 39
“However, there are still billions of known protein sequences waiting for structure determination, which is a huge task (Mitchell et al., 2020; Steinegger et al., 2019).”
BackgroundSun, X., Kong, T., Huang, D., Chen, Z., Zhang, Y., Häggblom, M. M., Soleimani, M., Liu, H., Ren, Y., Wang, Y., et al. (2024) · Environmental Science & Technology · doi:10.1021/acs.est.3c10945 · Cited by 37
Digging into the 3D Structure Predictions of AlphaFold2 with Low Confidence: Disorder and Beyond
Bruley, A., Mornon, J., Duprat, É., & Callebaut, I. (2022) · Biomolecules · doi:10.3390/biom12101467 · Cited by 37
“It is the most comprehensive database used by AF2 in order to build multiple sequence alignments, gathering sequences from genomic and metagenomic databases (UniprotKB [41] and metaclust [42] and datasets assembled with Plass [43]).”
MethodPakhrin, S. C., Pokharel, S., Pratyush, P., Chaudhari, M., Ismail, H. D., & KC, D. B. (2023) · Journal of Proteome Research · doi:10.1021/acs.jproteome.2c00667 · Cited by 37
Alarcón-Schumacher, T., Naor, A., Gophna, U., & Erdmann, S. (2022) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.2205037119 · Cited by 36
Malbranke, C., Bikard, D., Cocco, S., Monasson, R., & Tubiana, J. (2023) · Current Opinion in Structural Biology · doi:10.1016/j.sbi.2023.102571 · Cited by 40
Critical steps in an environmental metaproteomics workflow
Nebauer, D. J., Pearson, L. A., & Neilan, B. A. (2024) · Environmental Microbiology · doi:10.1111/1462-2920.16637 · Cited by 37
Novel opportunities for NGS-based one health surveillance of foodborne viruses
Desdouits, M., de Graaf, M., Strubbia, S., Munnink, B. B. O., Kroneman, A., Guyader, F. S. L., & Koopmans, M. (2020) · One Health Outlook · doi:10.1186/s42522-020-00015-6 · Cited by 35
“Recently, another strategy (Plass) proposes to convert reads into proteins before assembling into longer sequences, reducing variability and facilitating genomic identification [60].”
BackgroundCation and Anion Channelrhodopsins: Sequence Motifs and Taxonomic Distribution
Govorunova, E. G., Sineshchekov, O. A., Li, H., Wang, Y., Brown, L. S., Palmateer, A., Melkonian, M., Cheng, S., Carpenter, E. J., Patterson, J., et al. (2021) · mBio · doi:10.1128/mbio.01656-21 · Cited by 35
“We also analyzed 136 deep-ocean metatranscriptomic libraries from the TARA Oceans Expedition (64) assembled with the Plass protein-level algorithm (65).”
MethodPei, H., Li, J., Ma, S., Jiang, J., Li, M., Zou, Q., & Lv, Z. (2023) · Applied Sciences · doi:10.3390/app13052858 · Cited by 38
AutoPeptideML: a study on how to build more trustworthy peptide bioactivity predictors
Fernández-Díaz, R., Cossio-Pérez, R., Agoni, C., Lam, H. T., López, V., & Shields, D. C. (2024) · Bioinformatics · doi:10.1093/bioinformatics/btae555 · Cited by 35
“The models are trained on a vast set of sequences from the UniRef (Suzek et al. 2007) or BFD (Steinegger et al. 2019) databases which include not only protein sequences, but also peptides.”
MethodRecent Advances and Challenges in Protein Structure Prediction
Peng, C., Liang, F., Xia, Y., Zhao, K., Hou, M., & Zhang, G. (2023) · Journal of Chemical Information and Modeling · doi:10.1021/acs.jcim.3c01324 · Cited by 42
FusPB-ESM2: Fusion model of ProtBERT and ESM-2 for cell-penetrating peptide prediction
Zhang, F., Li, J., Wen, Z., & Fang, C. (2024) · Computational Biology and Chemistry · doi:10.1016/j.compbiolchem.2024.108098 · Cited by 35
Shen, T., Wu, J., Lan, H., Zheng, L., Pei, J., Wang, S., Liu, W., & Huang, J. (2021) · Proteins Structure Function and Bioinformatics · doi:10.1002/prot.26232 · Cited by 33
Binette, V., Mousseau, N., & Tufféry, P. (2022) · Journal of Chemical Theory and Computation · doi:10.1021/acs.jctc.1c01293 · Cited by 36
Improving AlphaFold2-based protein tertiary structure prediction with MULTICOM in CASP15
Liu, J., Guo, Z., Wu, T., Roy, R. S., Chen, C., & Cheng, J. (2023) · Communications Chemistry · doi:10.1038/s42004-023-00991-6 · Cited by 31
Improving protein tertiary structure prediction by deep learning and distance prediction in CASP14
Liu, J., Wu, T., Guo, Z., Hou, J., & Cheng, J. (2022) · Proteins: Structure, Function, and Bioinformatics · doi:10.1002/prot.26186 · Cited by 32
“For some hard targets, the MSAs generated by HHblits on the Big Fantastic Database (BFD) (Steinegger, Mirdita, & Söding, 2019; Steinegger & Söding, 2018) are also used to predict distance maps.”
MethodAleksenko, V. A., Anand, D., Remeeva, A., Nazarenko, V. V., Gordeliy, V., Jaeger, K., Krauß, U., & Gushchin, I. (2020) · Catalysts · doi:10.3390/catal10091072 · Cited by 33
“The set B7 contained the metagenomic sequences from Tara oceans [109,110] identified using the MMseqs2 webserver [111,112].”
MethodShen, J., Yu, Q., Chen, S., Tan, Q., Li, J., & Li, Y. (2023) · Nature Computational Science · doi:10.1038/s43588-023-00576-2 · Cited by 32
“To generate high-quality protein sequences, we first integrate fastp [36] as a quality controlling tool and utilize PLASS [37] to assemble nucleotide sequences to protein sequences, which leads to 40,391,320 sequences.”
Advancing bioinformatics with language models: components, applications, and perspectives
Liu, J., Yang, M., Yu, Y., Xu, H., Wang, T., Li, K., & Zhou, X. (2026) · Briefings in Bioinformatics · doi:10.1093/bib/bbag367 · Cited by 66
TransPTM: a transformer-based model for non-histone acetylation site prediction
Meng, L., Chen, X., Cheng, K., Chen, N., Zheng, Z., Wang, F., Sun, H., & Wong, K.-C. (2024) · Briefings in Bioinformatics · doi:10.1093/bib/bbae219 · Cited by 31
“It is initially trained on the Big Fantastic Database (BFD) [36], which contains 65 million protein families cataloged using multiple sequence alignments (MSAs) and hidden Markov models.”
MethodTemBERTure: advancing protein thermostability prediction with deep learning and attention mechanisms
Rodella, C., Lazaridi, S., & Lemmin, T. (2024) · Bioinformatics Advances · doi:10.1093/bioadv/vbae103 · Cited by 31
“It was built on top of the pre-trained protBERT-BFD model (Elnaggar et al. 2022), a BERT model composed of 30 layers, 16 heads, and 1024 hidden layers and trained on over 2 billion protein sequences from the BFD100 (Steinegger and S € oding 2018, Steinegger et al. 2019) dataset.”
Hao, Y., Liu, X., Fu, H., Shao, X., & Cai, W. (2024) · Bioinformatics · doi:10.1093/bioinformatics/btae497 · Cited by 32
“ProtT5 adopts a 24-layer transformer architecture with 1024 hidden layers size, pretrained on the Big Fantastic Database (BFD) (Steinegger and S € oding 2018, Steinegger et al. 2019) and subsequently fine-tuned on UniRef50 (Suzek et al. 2015).”
BackgroundGoing to extremes – a metagenomic journey into the dark matter of life
Aevarsson, A., Kaczorowska, A., Aðalsteinsson, B. Þ., Ahlqvist, J., Al‐Karadaghi, S., Altenbuchner, J., Arsın, H., Atlason, Ú. Á., Brandt, D., Cichowicz-Cieślak, M., et al. (2021) · FEMS Microbiology Letters · doi:10.1093/femsle/fnab067 · Cited by 33
Self-citationWang, X. R., Ding, Z., Wang, R., & Lin, X. (2023) · Briefings in Bioinformatics · doi:10.1093/bib/bbac631 · Cited by 33
Protein embedding based alignment
Iovino, B. G., & Ye, Y. (2024) · BMC Bioinformatics · doi:10.1186/s12859-024-05699-5 · Cited by 29
“This approach can be applied to protein sequences where language models are trained on large protein databases, like UniRef50 [21], UniRef100 [21], and BFD [20] [19], and are able to learn about the “language of life” [4].”
MethodSpatialPPI: Three-dimensional space protein-protein interaction prediction with AlphaFold Multimer
(2024) · doi:10.1016/j.csbj.2024.03.009 · Cited by 29
“The AlphaFold Multimer was executed using the jackhmmer search from HMMER3 [47] targeting Uniref90 [48], UniProt [49], and MGnify [50], and a joint search of the Big Fantastic Database [51,52] and UniRef30[53] using HHBlits [54].”
MethodChen, X., Huang, J., Shen, T., Zhang, H., Xu, L., Yang, M., Xie, X., Yan, Y., & Yan, J. (2024) · Bioinformatics · doi:10.1093/bioinformatics/btae319 · Cited by 29
“…structures of only 100 000 proteins are currently resolved, and there are billions of proteins and peptides with known sequences but unknown structures, as well as unknown active pocket sequences and positions (Steinegger et al. 2019, Mitchell et al. 2020), we designed two versions of the model.”
Methodproovframe: frameshift-correction for long-read (meta)genomicsPreprint
Hackl, T., Trigodet, F., Eren, A. M., Biller, S. J., Eppley, J. M., Luo, E., Burger, A. H., DeLong, E. F., & Fischer, M. (2021) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2021.08.23.457338 · Cited by 29
Ultra-fast protein structure prediction to capture effects of sequence variation in mutation moviesPreprint
Weißenow, K., Heinzinger, M., Steinegger, M., & Rost, B. (2022) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2022.11.14.516473 · Cited by 29
Self-citationProtein embeddings improve phage-host interaction prediction
Gonzales, M. E. M., Ureta, J. C., & Shrestha, A. M. S. (2023) · PLoS ONE · doi:10.1371/journal.pone.0289030 · Cited by 28
Lee, J.-W., Won, J.-H., Jeon, S., Choo, Y., Yeon, Y., Oh, J.-S., Kim, M., Kim, S., Joung, I., Jang, C., et al. (2023) · Bioinformatics · doi:10.1093/bioinformatics/btad712 · Cited by 29
“For the MSAs, DeepFold uses standard MSA generation procedure of AF2 using JackHMMER (Johnson et al. 2010) with MGnify (Richardson et al. 2023), JackHMMER with UniRef90 (Suzek et al. 2015), and HHblits (Remmert et al. 2012) with Uniclust30 (Mirdita et al. 2017)/BFD (Steinegger et al. 2019).”
Structural Comparison and Drug Screening of Spike Proteins of Ten SARS-CoV-2 Variants
Yang, Q., Jian, X., Syed, A. A. S., Fahira, A., Zheng, C., Zhu, Z., Wang, K., Zhang, J., Wen, Y., Li, Z., et al. (2022) · Research · doi:10.34133/2022/9781758 · Cited by 30
Sen, N., Anishchenko, I., Bordin, N., Sillitoe, I., Velankar, S., Baker, D., & Orengo, C. (2022) · Briefings in Bioinformatics · doi:10.1093/bib/bbac187 · Cited by 25
“…with additional homologs by performing iterative sequence hhblits [102] searches against uniclust30 (version UniRef30_2020_01) [103] and BFD [104, 105] databases as outlined by Anishchenko and colleagues [106] (details in Supplementary Text 12 available online at…”
Seq-InSite: sequence supersedes structure for protein interaction site prediction
Hosseini, S., Golding, G. B., & Ilie, L. (2024) · Bioinformatics · doi:10.1093/bioinformatics/btad738 · Cited by 26
Lee, K., Kim, D., & Cha, C. (2021) · The Journal of Microbiology · doi:10.1007/s12275-021-0652-4 · Cited by 26
Bioinformatics approaches for unveiling virus-host interactions
Iuchi, H., Kawasaki, J., Kubo, K., Fukunaga, T., Hokao, K., Yokoyama, G., Ichinose, A., Suga, K., & Hamada, M. (2023) · Computational and Structural Biotechnology Journal · doi:10.1016/j.csbj.2023.02.044 · Cited by 25
“plete viral genomes is also a high-priority issue, reflecting the increasing number of novel viruses being discovered by metagenomic analysis [60].”
Xu, S., & Onoda, A. (2023) · Journal of Chemical Information and Modeling · doi:10.1021/acs.jcim.3c01202 · Cited by 26
Streptomyces umbrella toxin particles block hyphal growth of competing species
Zhao, Q., Bertolli, S. K., Park, Y., Tan, Y., Cutler, K. J., Srinivas, P., Asfahl, K. L., Garcia, C. F., Gallagher, L. A., Li, Y., et al. (2024) · Nature · doi:10.1038/s41586-024-07298-z · Cited by 26
“In brief, MSAs were generated for UmbB1, UmbB2, UmbB3, UmbC1, UmbC2 and UmbC3 by running HHblits 34 against Uniref30 and BFD, and paired MSAs for all three pairs were generated by maxing taxonomy identifiers.”
Yao, L., Xie, P., Guan, J., Chung, C., Zhang, W., Deng, J., Huang, Y., Chiang, Y., & Lee, T.-Y. (2024) · Briefings in Bioinformatics · doi:10.1093/bib/bbae460 · Cited by 27
“ProtBert is pretrained using the BFD dataset [35],which encompasses 393 billion amino acids from 2.1 billion protein sequences.”
InfluentialHeinzinger, M., & Rost, B. (2025) · Current Opinion in Structural Biology · doi:10.1016/j.sbi.2025.102986 · Cited by 27
Liu, J., Zhao, K., & Zhang, G. (2023) · Briefings in Bioinformatics · doi:10.1093/bib/bbac507 · Cited by 25
DIPS-Plus: The enhanced database of interacting protein structures for interface prediction
Morehead, A., Chen, C., Sedova, A., & Cheng, J. (2023) · Scientific Data · doi:10.1038/s41597-023-02409-3 · Cited by 26
“Each HMM profile used to ascertain these residue-specific transition and emission probabilities are constructed by HHmake [35] using MSAs that were generated after two iterations by HHblits [35] and the Big Fantastic Database (BFD) (version: March 2019) of protein sequences [36].”
MethodLi, Y., Xu, X., Zhang, X., Xu, Z., Zhao, J., Zhu, R., Wang, Z., Ran, W., Zhao, W., Yan, N., et al. (2025) · Gut Microbes · doi:10.1080/19490976.2025.2523811 · Cited by 28
Gamouh, H., Novotný, M., & Hoksza, D. (2025) · Bioinformatics · doi:10.1093/bioinformatics/btaf431 · Cited by 23
Target identification and assessment in the era of AI
Pun, F. W., Podolskiy, D. I., Izumchenko, E., Mortlock, A. A., Oprea, T. I., Scheibye‐Knudsen, M., Fortney, K., Morgen, E. K., Ren, F., & Zhavoronkov, A. A. (2026) · Nature Reviews Drug Discovery · doi:10.1038/s41573-026-01412-8 · Cited by 31
“Proteins are flexible and dynamic 150 , with experimentally determined structures available for only a small fraction of the proteome 151,152 .”
Li, Y., Zhang, C., Zheng, W. H., Zhou, X., Bell, E. W., Yu, D., & Zhang, Y. (2021) · Proteins Structure Function and Bioinformatics · doi:10.1002/prot.26211 · Cited by 24
Sensitive clustering of protein sequences at tree-of-life scale using DIAMOND DeepClustPreprint
Buchfink, B., Ashkenazy, H., Reuter, K., Kennedy, J. A., & Drost, H. (2023) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2023.01.24.525373 · Cited by 23
Gonzales, M. E. M., Ureta, J. C., & Shrestha, A. M. S. (2024) · Bioinformatics · doi:10.1093/bioinformatics/btaf016 · Cited by 23
“These models were pretrained in a self-supervised fashion on sequences from large-scale protein databases [6, 55, 50, 52] and have been shown to cap-43 ture relevant physicochemical characteristics [21, 47, 27, 40].”
MethodEvaluating large language models for annotating proteins
Vitale, R., Bugnon, L. A., Fenoy, E., Milone, D. H., & Stegmayer, G. S. (2024) · Briefings in Bioinformatics · doi:10.1093/bib/bbae177 · Cited by 22
“Two auto-regressive models (Transformer-XL and XLNet) and four auto-encoder models (BERT, Albert, Electra and T5) were trained with more than 2000 million proteins and up to 393 billion amino acids from UniRef50 and BFD databases [32].”
MethodEvolution of a chordate-specific mechanism for myoblast fusion
Zhang, H., Shang, R., Kim, K., Zheng, W. H., Johnson, C. J., Sun, L., Niu, X., Liu, L., Zhou, J., Liu, L., et al. (2022) · Science Advances · doi:10.1126/sciadv.add2696 · Cited by 21
Mesdaghi, S., Murphy, D., Rodríguez, F. S., Burgos-Mármol, J. J., & Rigden, D. J. (2020) · F1000Research · doi:10.12688/f1000research.27676.1 · Cited by 23
Baltoumas, F. A., Karatzas, E., Páez-Espino, D., Venetsianou, N. K., Aplakidou, E., Oulas, A., Finn, R. D., Ovchinnikov, S., Pafilis, E., Kyrpides, N. C., et al. (2023) · Frontiers in Bioinformatics · doi:10.3389/fbinf.2023.1157956 · Cited by 21
“Short-read metagenomic assembly software includes tools such as metaSPAdes (Nurk et al., 2017), MetaviralSPAdes (a variant of the former for viral metagenomes) (Antipov et al., 2020), Plass (Steinegger et al., 2019b), MEGAHIT (Li et al., 2015), MetaVelvet (Namiki et al., 2012), Omega (Haider et al., 2014), Ray Meta (B…”
MethodRossier, V., Warwick Vesztrocy, A., Robinson-Rechavi, M., & Dessimoz, C. (2021) · Bioinformatics · doi:10.1093/bioinformatics/btab219 · Cited by 25
Structural analogue-based protein structure domain assembly assisted by deep learning
Peng, C.-X., Zhou, X.-G., Xia, Y.-H., Liu, J., Hou, M.-H., & Zhang, G.-J. (2022) · Bioinformatics · doi:10.1093/bioinformatics/btac553 · Cited by 21
Insights into the inner workings of transformer models for protein function prediction
Wenzel, M. A., Grüner, E., & Strodthoff, N. (2024) · Bioinformatics · doi:10.1093/bioinformatics/btae031 · Cited by 24
“This Supplementary material also explains the finetuning of the transformers “ProtBert-BFD” and “ProtT5-XL-UniRef50” (Elnaggar et al. 2022) and “ESM-2” (Lin et al. 2023) on the GO and EC tasks, and contains statements about data availability and composition.”
BackgroundAn Expanded Gene Catalog of Mouse Gut Metagenomes
Zhu, J., Ren, H., Zhong, H., Li, X., Zou, Y., Han, M., Li, M., Madsen, L., Kristiansen, K., & Xiao, L. (2021) · mSphere · doi:10.1128/msphere.01119-20 · Cited by 21
PredictProtein – Predicting Protein Structure and Function for 29 YearsPreprint
Bernhofer, M., Dallago, C., Karl, T., Satagopam, V., Heinzinger, M., Littmann, M., Olenyi, T., Qiu, J., Schütze, K., Yachdav, G., et al. (2021) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2021.02.23.432527 · Cited by 21
Self-citationPlastic biodegradation by sediment microbial populations under denitrifying conditions
Sun, X., Wang, S., Lin, Z., Chen, Z., Huang, W., Kong, T., Huang, D., Li, B., Zhang, H., & Sun, W. (2025) · Environmental Science & Technology · doi:10.1021/acs.est.4c12187 · Cited by 20
Assessing the role of evolutionary information for enhancing protein language model embeddings
Erckert, K., & Rost, B. (2024) · Scientific Reports · doi:10.1038/s41598-024-71783-8 · Cited by 21
“A pLM trained on large databases like BFD 69 may implicitly learn the same constraints captured in an MSA simply by encountering sequences from larger families more frequently.”
Deep learning geometrical potential for high-accuracy ab initio protein structure prediction
Li, Y., Zhang, C., Yu, D., & Zhang, Y. (2022) · iScience · doi:10.1016/j.isci.2022.104425 · Cited by 20
“HH-suite Steinegger et al., 2019a https://github.com/soedinglab/hh-suite HMMER Eddy, 1998 http://hmmer.org TM-score Zhang and Skolnick, 2004b https://zhanggroup.org/TM-score CCMPred Seemayer et al., 2014 https://github.com/soedinglab/CCMpred TripletRes Li et al., 2021a https://zhanggroup.org/TripletRes CD-HIT Fu et al…”
BackgroundMethodInfluentialSugarcane cystatins: From discovery to biotechnological applications
Shibao, P. Y. T., Júnior, C. D. S., Santiago, A. C., Mohan, C., Miguel, M. C., Toyama, D., Vieira, M. A. S., Narayanan, S., Figueira, A., Carmona, A. K., et al. (2020) · International Journal of Biological Macromolecules · doi:10.1016/j.ijbiomac.2020.11.185 · Cited by 20
LambdaPP : Fast and accessible protein‐specific phenotype predictions
(2022) · doi:10.1002/pro.4524 · Cited by 19
Self-citationManfredi, M., Savojardo, C., Martelli, P. L., & Casadio, R. (2023) · Journal of Molecular Biology · doi:10.1016/j.jmb.2023.167963 · Cited by 20
Zhang, L., & Liu, T. (2024) · International Journal of Biological Macromolecules · doi:10.1016/j.ijbiomac.2024.136147 · Cited by 19
A-Prot: protein structure modeling using MSA transformer
Hong, Y., Lee, J., & Ko, J. (2022) · BMC Bioinformatics · doi:10.1186/s12859-022-04628-8 · Cited by 19
“Through the token and position embedding of the MSA Transformer, the input MSA is embedded into a r × c × 768 tensor which is the input and output of each attention block.”
MethodPredicting Mutation-Disease Associations Through Protein Interactions Via Deep Learning
Li, X., Cao, B., Wang, J., Meng, X., Wang, S., Huang, Y., Petretto, E. G., & Song, T. (2025) · IEEE Journal of Biomedical and Health Informatics · doi:10.1109/jbhi.2025.3541848 · Cited by 18
“Specifically, this model is based on the general BERT framework and completes masked language modeling and next sentence prediction tasks on a corpus containing 2.1 billion protein sequences [43], ultimately achieving the mapping of input protein sequences to high-dimensional feature vectors.”
InfluentialAssessing protein model quality based on deep graph coupled networks using protein language model
(2023) · doi:10.1093/bib/bbad420 · Cited by 18
“Finally, the graph coupled network outputs the results of evaluating the model. produced through HHbits [31] searching against UniRef30 [32] and BFD [33], and then the searched MSA is fed into the ESM-MSA-1b [34] language model to derive residue-level sequence embeddings (768-dim) and row-attention…”
MethodWu, J., Liu, Y., Ge, F., & Yu, D. (2024) · Computers in Biology and Medicine · doi:10.1016/j.compbiomed.2024.108227 · Cited by 19
PRESCOTT: a population aware, epistatic, and structural model accurately predicts missense effects
Tekpinar, M., David, L., Henry, T., & Carbone, A. (2025) · Genome Biology · doi:10.1186/s13059-025-03581-y · Cited by 18
“Since Alphafold implementation uses several databases such as Uniref90 [68], BFD [69, 70], and Mgnify [71], due to their large sizes, Alphafold computational time becomes prohibitively high for thousands of proteins and, for this reason, we used Colabfold, based on highly clustered databases, to…”
End-to-end multitask learning, from protein language to protein features without alignmentsPreprint
Elnaggar, A., Heinzinger, M., Dallago, C., & Rost, B. (2019) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/864405 · Cited by 17
Advancing protein structure prediction beyond AlphaFold2
Park, S., Myung, S., & Baek, M. (2025) · Current Opinion in Structural Biology · doi:10.1016/j.sbi.2025.102985 · Cited by 18
Recent Advances in Protein Homology Detection Propelled by Inter-Residue Interaction Map Threading
Bhattacharya, S., Roche, R., Shuvo, M. H., & Bhattacharya, D. (2021) · Frontiers in Molecular Biosciences · doi:10.3389/fmolb.2021.643752 · Cited by 18
“For example, BFD (Steinegger et al., 2019), a recent sequence database, is one of the largest sequence databases containing 2 billion protein sequences from soil samples and 292 million sequences of marine samples.”
BackgroundOda, T. (2023) · Proteins: Structure, Function, and Bioinformatics · doi:10.1002/prot.26551 · Cited by 17
Deciphering the protein landscape with ProtFlash, a lightweight language model
Wang, L., Zhang, H., Xu, W. M., Xue, Z., & Wang, Y. (2023) · Cell Reports Physical Science · doi:10.1016/j.xcrp.2023.101600 · Cited by 20
Palacios, A. V., Acharya, P., Peidl, A. S., Beck, M. R., Blanco, E., Mishra, A., Bawa‐Khalfe, T., & Pakhrin, S. C. (2024) · NAR Genomics and Bioinformatics · doi:10.1093/nargab/lqae011 · Cited by 17
“The PLM ProtT5 was trained on unlabeled protein sequences from BFD (Big Fantastic Database; 2.5 billion sequences including meta-genomic sequences) ( 66 ), and UniRef50 ( 67 ).”
MethodInfluentialSun, X., Wu, Z., Su, J., & Li, C. (2024) · International Journal of Biological Macromolecules · doi:10.1016/j.ijbiomac.2024.136933 · Cited by 18
Empathi: embedding-based phage protein annotation tool by hierarchical assignment
Boulay, A., Leprince, A., Enault, F., Rousseau, E., & Galiez, C. (2025) · Nature Communications · doi:10.1038/s41467-025-64177-5 · Cited by 16
Improving deep learning-based protein distance prediction in CASP14
Guo, Z., Wu, T., Liu, J., Hou, J., & Cheng, J. (2021) · Bioinformatics · doi:10.1093/bioinformatics/btab355 · Cited by 19
AttSec: protein secondary structure prediction by capturing local patterns from attention map
(2023) · doi:10.1186/s12859-023-05310-3 · Cited by 16
“ProtT5-XL-U50, which is based on T5 [15], is trained using the BFD dataset [16, 17] and the UniRef50 dataset [18] by performing a denoising task proposed in BERT [19] as a pretext task.”
MethodApplication of Sequence Embedding in Protein Sequence-Based Predictions
Ibtehaz, N., & Kihara, D. (2022) · WORLD SCIENTIFIC eBooks · doi:10.1142/9789811258589_0002 · Cited by 18
Petascale Homology Search for Structure PredictionPreprint
Lee, S., Kim, G., Karin, E. L., Mirdita, M., Park, S., Chikhi, R., Babaian, A., Kryshtafovych, A., & Steinegger, M. (2023) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2023.07.10.548308 · Cited by 14
“We used HHmake (Steinegger et al. 2019a) to compute the number of effective sequences (N eff ), where higher values indicate less similarity between the sequences and more diverse MSAs.”
MethodSelf-citationFast and accurate modeling and design of antibody-antigen complex using tFoldPreprint
Wu, F., Zhao, Y. S., Wu, J., Jiang, B., He, B., Huang, L.-K., Qin, C., Yang, F., Huang, N., Xiao, Y., et al. (2024) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2024.02.05.578892 · Cited by 16
The claim of primacy of human gut Bacteroides ovatus in dietary cellobiose degradation
Li, M., Wang, Y., Guo, C., Wang, S., Zheng, L., Bu, Y., & Ding, K. (2023) · Gut Microbes · doi:10.1080/19490976.2023.2227434 · Cited by 16
“Following, the monomer structure was predicted based on the inference of Alphafold2 to construct the 3D model and energy optimized.(58) After the predication of protein structure was finished, the possible small molecular binding site and amino acids were predicted based on the point-site.”
MethodMachine learning for predicting protein properties: A comprehensive review
Wang, Y., Zhang, Y., Zhan, X., He, Y., Yang, Y., Cheng, L., & Alghazzawi, D. (2024) · Neurocomputing · doi:10.1016/j.neucom.2024.128103 · Cited by 15
Pakhrin, S. C., Chauhan, N., Khan, S., Upadhyaya, J., Beck, M. R., & Blanco, E. (2024) · Bioinformatics · doi:10.1093/bioinformatics/btae643 · Cited by 15
Unifying the known and unknown microbial coding sequence spacePreprint
Vanni, C., Schechter, M. S., Acinas, S. G., Barberán, A., Buttigieg, P. L., Casamayor, E. O., Delmont, T. O., Duarte, C. M., Eren, A. M., Finn, R. D., et al. (2020) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2020.06.30.180448 · Cited by 15
Self-citationGenerating novel protein sequences using Gibbs sampling of masked language modelsPreprint
Johnson, S. R., Monaco, S., Massie, K., & Syed, Z. (2021) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2021.01.26.428322 · Cited by 18
“Pre-trained ESM12 produced sequences of much poorer quality than ESM34 and ProtBERT-BFD, but fine-tuning ESM12 allowed generation of sequences whose quality met or surpassed the quality of sequences generated from larger models.”
BackgroundInfluentialRareFold: Structure Prediction and Design of Proteins with Noncanonical Amino AcidsPreprint
Li, Q., Daumiller, D., Zuo, F., Marcotte, H., Pan-Hammarström, Q., & Bryant, P. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.05.19.654846 · Cited by 19
“This process constitutes the main bottleneck for the predictions as very large databases such as the Big Fantastic Database [26,27] are searched, which is very time-consuming [28].”
Using metagenomic data to boost protein structure prediction and discovery
Hou, Q., Pucci, F., Pan, F., Xue, F., Rooman, M. J., & Feng, Q. (2022) · Computational and Structural Biotechnology Journal · doi:10.1016/j.csbj.2021.12.030 · Cited by 18
“We start by reviewing the widely known and curated metagen-ome resources and databases: IMG/M [29], MGnify [30], MetaClust [31] and BFD [32].”
MethodWang, Z., Meng, J., Dai, Q., Li, H., Xia, S., Yang, R., & Luan, Y. (2024) · Expert Systems with Applications · doi:10.1016/j.eswa.2024.124168 · Cited by 14
Hu, J., Chen, K., Rao, B., Ni, J.-Y., Thafar, M. A., Albaradei, S., & Arif, M. (2024) · Analytical Biochemistry · doi:10.1016/j.ab.2024.115637 · Cited by 14
Kabir, A., Moldwin, A., Bromberg, Y., & Shehu, A. (2024) · Bioinformatics Advances · doi:10.1093/bioadv/vbae119 · Cited by 13
“ESM2 (Lin et (Steinegger et al. 2019, Steinegger and S € oding 2018) merged with UniProt and proteins translated from multiple metagenomic sequencing projects, making it the largest collection of protein sequences available at the time of writing even after removal of duplicates from the original…”
InfluentialLiu, J., Neupane, P., & Cheng, J. (2025) · Proteins Structure Function and Bioinformatics · doi:10.1002/prot.26850 · Cited by 18
“Sequence-based MSAs are constructed by searching the sequence of each subunit against a comprehensive set of sequence databases, including UniRef30 [16], UniRef90 [17], Metaclust [18], BFD [18, 19], Mgnify [20], UniProt [17], JGI [21], TaraDB [22], and MetaSourceDB [23].”
Large language models transform biological research: from architecture to utilization
Wang, T., & Luo, Z. (2025) · Science China Information Sciences · doi:10.1007/s11432-024-4466-3 · Cited by 14
ProteinGLUE multi-task benchmark suite for self-supervised protein modeling
Capel, H., Weiler, R., Dijkstra, M., Vleugels, R., Bloem, P., & Feenstra, K. A. (2022) · Scientific Reports · doi:10.1038/s41598-022-19608-4 · Cited by 11
“…sequences 42 , UniRep is trained on 24 million UniRef50 sequences, ESM-1b is pre-trained on 250 million UniParc sequences, and Elnaggar et al. 11 even train their models on the BFD dataset containing over 2 billion sequences 43,44 , and on the UniRef100 database with 216 million sequences.”
MethodLanguage models for protein design
Lee, J. S., Abdin, O., & Kim, P. M. (2025) · Current Opinion in Structural Biology · doi:10.1016/j.sbi.2025.103027 · Cited by 15
Nature‐inspired hierarchical materials
Fu, Q., Baino, F., Saiz, E., Bai, H., & Mauro, J. C. (2025) · Journal of the American Ceramic Society · doi:10.1111/jace.70156 · Cited by 17
“289 The breakthrough in de novo protein design through computational modeling and its validation through experimental synthesis by David Baker and his team demonstrated for the first time the possibility of creating entirely new proteins with specific structures and functions that do not exist in…”
High-accuracy protein complex structure modeling based on sequence-derived structure complementarity
Hou, M., Xia, Y., Wang, P., Lv, Z., Hou, D., Zhou, X., Zeng, J., & Zhang, G. (2025) · Nature Communications · doi:10.1038/s41467-025-65090-7 · Cited by 15
“Our method participated in CASP16 under the group name Guijunlab-Complex (Group #148) and Guijunlab-Assembly (Group #312) .”
Zhang, L., & Liu, T. (2024) · International Journal of Biological Macromolecules · doi:10.1016/j.ijbiomac.2024.135762 · Cited by 14
Metadata
- Authors
- Steinegger, M., Mirdita, M., & Söding, J.
- Affiliations
- Seoul National University (KR)
- Max Planck Institute for Biophysical Chemistry (DE)
- Corresponding author
- Steinegger, Martin
- Söding, Johannes
- Venue
- Nature Methods
- Published
- 2019-06-24
- Volume
- 16
- Issue
- 7
- Pages
- 603-606
- ISSN
- 1548-7091 (ISSN-L)
- 1548-7105
- OpenAlex
- W2949342052
- Cited by
- 445
- References
- 31
- Citations by year
- 2026: 38 · 2025: 78 · 2024: 101 · 2023: 78 · 2022: 71 · 2021: 50 · 2020: 23 · 2019: 6
- FWCI
- 12.42
- Type
- journal article
- Language
- en
- Open access
- hybrid
- doi.org
- Locations
- Nature Methods (publishedVersion, cc-by)
- bioRxiv (Cold Spring Harbor Laboratory) (acceptedVersion, cc-by-nc-nd)
- PubMed (publishedVersion)
- MPG.PuRe (Max Planck Society) (submittedVersion)
- Seoul National University Open Repository (Seoul National University) (submittedVersion)
- Funders
- Materials and Energy Research Center
- Topics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Genomics and Phylogenetic Studies
- Physical Sciences › Environmental Science › Ecology › Environmental DNA in Biodiversity Studies
- Physical Sciences › Environmental Science › Ecology › Microbial Community Ecology and Physiology
- Keywords
- metagenomics, metatranscriptomics, soil metagenome
- MeSH terms
- Amino Acid Sequence, Codon, Proteins, Open Reading Frames, Metagenomics
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
