[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-118065-en":3,"doc-seo-118065-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},118065,1099523882367,"Hazel","https://ap-avatar.wpscdn.com/davatar_9964176cb1d06d4a9deccf72a44ae3dc",8,"Research & Report","Machine learning sheds light on microbial dark proteins - Genome Watch","Genome Watch highlights recent machine-learning approaches that model atomic-level protein structures and help interpret microbial “dark” proteins lacking recognizable homologs or reliable functional annotations. Using predicted structures from resources such as AlphaFold/AFDB, Foldseek clustering, and metagenomic resources, the studies reduced the protein search space, discovered thousands of new protein clusters, and supported inference of family associations. Results also show how structural similarity can guide functional hypotheses when sequence-only methods fail.","Lawrence Berkeley National Laboratory LBL Publications  \nTitle  \nMachine learning sheds light on microbial dark proteins  \nPermalink  \n[https://escholarship.org/uc/item/3xs0q8x1](https://escholarship.org/uc/item/3xs0q8x1)  \nJournal  \nNature Reviews Microbiology, 22(2)  \nISSN  \n1740-1526  \nAuthors  \nHammack, Aeron Tynes Blaby-Haas, Crysten E  \nPublication Date  \n2024-02-01  \nDOI  \n10.1038/s41579-023-01002-0  \nPeer reviewed  \n[eScholarship.org](eScholarship.org) Powered by the California Digital Library  \nUniversity of California  \nGENOME WATCH  \nMachine learning sheds light on microbial dark proteins  \nAeron Tynes Hammack and Crysten E. Blaby-Haas  \n5 This month’s Genome Watch highlights Two of these studies analysed 215 mil- 95 families (NMPFs), were idnetified. The use  \nthe recent use of machine learning to un- 50 lion precomputed structures in the Al- of AlphaFold and the clustering of NMPFs  \ncover functional ‘dark matter’ in the mi- phaFold Database (AFDB)3,4. One of the based on structure resulted in ~4,000  \n[crobial protein universe.](crobial protein universe. works3 developed)[ works](crobial protein universe. works3 developed)[3](crobial protein universe. works3 developed)[ developed](crobial protein universe. works3 developed) a method, Foldseek clus- unique predicted structures. Although not  \nter, which uses a combination of ultra-fast apparent at the sequence level, structural  \nMetagenomics projects have revealed sequence and structural aligners to cluster 100 similarity placed 62% of protein structures  \n0 more than 8 billion non-redundant micro- 55 sequences and then cluster representative in a known family.  \nbial protein sequences from across the structures. After quality filtering, this ap- In sum, these studies identified new  \nEarth’s biosphere1. Of these, 1.17 billion proach reduced the AFDB protein space to protein families and demonstrated the  \nproteins do not have recognizable homo- 2. 3 million structures. Of these, a little over value of structural similarity in identifying logues in any of the more than 100,000 ref- 700,000 protein clusters (~30%) do not 105 family association, especially for highly di-  \n5 erence genomes available1. Understanding 60 have matches to experimentally deter- vergent sequences. Although defining the the function of these microbial proteins is a mined structures and could not be func- structure of an uncharacterized protein daunting task. Fortunately, machine learn- tionally annotated with Pfam or TIGRFAM does not necessarily reveal its function, ing (ML), has recently achieved unprece- annotations. However, in several cases, structural similarity to characterized prodented accuracy in modelling complex bio- structural similarity to annotated clusters, 110 teins can provide an invaluable inference  \n0 logical data and making predictions. At the 65 including leveraging human proteins to in- when seeking to decode the vast functional forefront of these advancements are ML- form on bacterial proteins, enabled func- information contained within microbial gebased approaches that can confidently pre- tional predictions for several bacterialpro- nomes.  \ndict atomic-level protein structures for teins in ‘dark’ —that is, poorly annotated— Aeron Tynes Hammack1 and Crysten E. Blaby-Haas1,2*  \nmany (but not all) amino acid sequences. . clusters. 115 1Molecular Foundry, Lawrence Berkeley National Laboratory,  \n5 A recent study used the ESMFold predic- 70 The other study4 used precomputed Berkeley, CA; 2DOE Joint Genome Institute, Lawrence Berkeley  \ntor2 that takes advantage of a large lan- clusters from the UniProt database to de- National Laborato*e-ry,mBeail:rkceblley,abyCA@lb, lUSAgov.  \nguage model (LLM) to quickly generate fine a set of 6 million representative struc-  \n617 million structures from the European tures. These sequences were then used to [https://doi.org/](https://doi.org/)10.1038/s41579-XXX-XXXX-X Bioinformatics Institute (EBI)’s MGnify build an interactive sequ","cbCaicYJLt7umwXy","https://ap.wps.com/l/cbCaicYJLt7umwXy","pdf",306103,1,2,"English","en",105,"# Genome Watch\n## Machine learning for atomic-level protein prediction\n## Clustering and structural similarity inference\n## Case studies from metagenomics and protein databases","[{\"question\":\"What are “microbial dark proteins” in this Genome Watch?\",\"answer\":\"They refer to microbial protein sequences that lack recognizable homologues in reference genomes and are poorly annotated, making functional interpretation difficult.\"},{\"question\":\"How does machine learning help interpret dark protein function?\",\"answer\":\"By confidently predicting atomic-level protein structures and then using structural similarity and clustering to infer family association and functional context even when sequence similarity is limited.\"},{\"question\":\"Which resources or methods are used to predict and cluster protein structures?\",\"answer\":\"The highlights include AlphaFold-based structure prediction in AFDB, Foldseek clustering, and precomputed representative structures derived from UniProt and metagenomic resources such as MGnify and IMG/M.\"}]","Machine learning sheds light on microbial dark proteins - Genome Watch | PDF",1785681183,5,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"machine-learning-sheds-light-on-microbial-dark-proteins-genome-watch","",{"@graph":36,"@context":85},[37,53,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,47,50],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":21},"https://docshare.wps.com/document/","Document",{"item":48,"name":12,"@type":43,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":43,"position":52},"https://docshare.wps.com/document/machine-learning-sheds-light-on-microbial-dark-proteins-genome-watch/118065/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":41,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-09-05","2026-08-02",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"What are “microbial dark proteins” in this Genome Watch?","Question",{"text":75,"@type":76},"They refer to microbial protein sequences that lack recognizable homologues in reference genomes and are poorly annotated, making functional interpretation difficult.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does machine learning help interpret dark protein function?",{"text":80,"@type":76},"By confidently predicting atomic-level protein structures and then using structural similarity and clustering to infer family association and functional context even when sequence similarity is limited.",{"name":82,"@type":73,"acceptedAnswer":83},"Which resources or methods are used to predict and cluster protein structures?",{"text":84,"@type":76},"The highlights include AlphaFold-based structure prediction in AFDB, Foldseek clustering, and precomputed representative structures derived from UniProt and metagenomic resources such as MGnify and IMG/M.","https://schema.org",{"og:url":51,"og:type":87,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":89,"canonical":51},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,109,114,119,122,127,130,134],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":21,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":29,"doc_module":4,"doc_module_name":46,"category_name":106,"show_sort_weight":107,"slug":108},"Comic",60,"comic",{"id":110,"doc_module":4,"doc_module_name":46,"category_name":111,"show_sort_weight":112,"slug":113},6,"Technology",50,"technology",{"id":115,"doc_module":4,"doc_module_name":46,"category_name":116,"show_sort_weight":117,"slug":118},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":120,"slug":121},30,"research-report",{"id":123,"doc_module":4,"doc_module_name":46,"category_name":124,"show_sort_weight":125,"slug":126},9,"Religion & Spirituality",20,"religion-spirituality",{"id":125,"doc_module":4,"doc_module_name":46,"category_name":128,"show_sort_weight":125,"slug":129},"World Cup","world-cup",{"id":131,"doc_module":4,"doc_module_name":46,"category_name":132,"show_sort_weight":131,"slug":133},10,"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":46,"category_name":136,"show_sort_weight":29,"slug":137},19,"General","general"]