CoChat
Sign Up
Cited by 4,602Genomics and Phylogenetic StudiesProtein Structure and DynamicsMachine Learning in Bioinformatics

The Pfam protein families database in 2019

Open access
El-Gebali, Sara, Mistry, Jaina, Bateman, Alex, Eddy, Sean R, Luciani, Aurélien, Potter, Simon C, Qureshi, Matloob, Richardson, Lorna J +8 more·2019·Nucleic Acids Research
DOI 10.1093/nar/gky995PMC PMC6324024
Open this paper in CoChat — free accountOpen PDF

Abstract

The last few years have witnessed significant changes in Pfam (https://pfam.xfam.org). The number of families has grown substantially to a total of 17,929 in release 32.0. New additions have been coupled with efforts to improve existing families, including refinement of domain boundaries, their classification into Pfam clans, as well as their functional annotation. We recently began to collaborate with the RepeatsDB resource to improve the definition of tandem repeat families within Pfam. We carried out a significant comparison to the structural classification database, namely the Evolutionary Classification of Protein Domains (ECOD) that led to the creation of 825 new families based on their set of uncharacterized families (EUFs). Furthermore, we also connected Pfam entries to the Sequence Ontology (SO) through mapping of the Pfam type definitions to SO terms. Since Pfam has many community contributors, we recently enabled the linking between authorship of all Pfam entries with the corresponding authors' ORCID identifiers. This effectively permits authors to claim credit for their Pfam curation and link them to their ORCID record.

AI snapshot

Written from the paper’s full text.

TL;DR

Pfam 32.0 (September 2018) contains 17,929 protein family entries and reports coverage of 77.2% of UniProtKB sequences and 53.2% of residues. The release adds 1,664 new entries, 825 generated via comparison with ECOD, expands clans to 628, and improves repeat definitions, DUF annotation, SO mapping, and ORCID-based authorship credit.

Key findings

  • Pfam 32.0 has 17,929 entries; UniProtKB coverage is 77.2% of sequences and 53.2% of residues, while reference proteome coverage is 74.5% of sequences and 50.1% of residues.
  • The release adds 1,664 new entries; 825 new entries (50%) were generated using ECOD, and over 400 existing entries changed domain boundaries between Pfam 29.0 and 32.0. Pfam 32.0 covers 87.1% of PDB sequences and 73.8% of PDB residues.
  • There are 628 clans total, with 74 new clans since Pfam 29.0; 7,001 entries belong to clans, corresponding to 74% of all sequence regions annotated by Pfam.
  • 272 DUF entries were annotated and renamed, but 3,961 DUFs remain (22% of all entries); 39 new entries were deposited by RepeatsDB curators. Of 313 unique Pfam authors, 161 have an ORCID, and 97% of Pfam entries have at least one author with an ORCID.

Methods

  • Database release/update: Pfam 32.0 was released in September 2018; entries are built from seed alignments and profile HMMs queried against pfamseq, which is derived from UniProtKB Reference Proteomes.
  • Comparisons and integration: Pfam 31.0 was compared with ECOD version 29 to incorporate ECOD Unclassified Families (EUFs); SCOOP all-against-all analysis identified additional clan relationships; RepeatsDB collaboration refined repeat models; Sequence Ontology terms were mapped to Pfam types; ORCID identifiers were linked to author lines.
  • Analysis: HMMER searches, E-value thresholds (<0.001 for ECOD EUF matches), curator-driven iterative searching, overlap resolution, and clan assignment were used to add and refine entries.

Limitations

  • The paper states that nearly 25% of sequences in pfamseq are pending annotation, and that 3,961 DUFs (22% of entries) remain, with an estimate that over a quarter of Pfam entries lack an experimentally validated function.
  • The paper states that no single repeat-identification approach ensures correct periodicity or full representation of tandem repeats, and repeat boundary definition remains challenging.
  • Evident from the text: this is a descriptive database update rather than a controlled benchmark; coverage changes are reported against expanding sequence and structure databases, and the marginal PDB coverage increase (<1%) is contextualized by 37% PDB growth.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1093/nar/gky995

Cites

25 of 25 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

0 of 13 cited sources checked · 0 supported · 0 flagged

  • Findingown data

    Using the Simple Comparison Of Outputs Program (SCOOP), an all-against-all analysis identified an additional 22 Pfam entries which were consequently added to a clan.

    “For example, using the Simple Comparison Of Outputs Program (12), we performed a comprehensive “all-against-all” analysis that identified an additional 22 Pfam entries which were consequently added to a clan.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [12] Bateman, 2007, SCOOP: a simple method for identification of novel protein superfamily relationships, Bioinformatics

      Not checked in the free run.

  • Findingown data

    Residue coverage of the reference proteomes in pfamseq significantly increased since Pfam 29.0.

    “On the other hand, the residue coverage of the reference proteomes in pfamseq significantly increased since Pfam 29.0.”
    Quote in the paperNot a substantial claim
  • Findingown data

    UniProtKB increased by 65 million sequences, a growth of 128%, between Pfam 29.0 and 32.0.

    “UniProtKB concurrently increased by 65 million sequences, a growth of 128%.”
    Quote in the paperNot a substantial claim
  • Findingnone given

    39 new entries have been deposited in Pfam by the RepeatsDB curators.

    “Thus far, 39 new entries have been deposited in Pfam by the RepeatsDB curators.”
    Quote in the paperNot a substantial claim
  • Findingown data

    DUF5328 (Pfam: PF17262) from Pfam 31.0 was identified as corresponding to two ECOD domains.

    “For example, Pfam entry DUF5328 (a ‘domain of unknown function’ [DUF], Pfam: PF17262) from Pfam 31.0 was identified as corresponding to two ECOD domains.”
    Quote in the paperNot a substantial claim
  • May combine two claims. 21
  • Classified as probably not a substantial claim. 12

  • May combine two claims.

    Findingown data

    The current Pfam 32.0 release contains 3,961 DUFs, corresponding to 22% of all entries.

    “The current Pfam 32.0 release contains 3,961 DUFs (corresponding to 22% of all entries), in addition to entries that have not been labelled as DUFs, since they are referred to by common names derived from literature [e.g. HCMV_UL124 (Pfam: PF17609) is a family of viral membrane glycoproteins of unknown function].”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingreasoning

    The Hva1_TUDOR entry was also added to the Tudor domain clan (CL0049).

    “Furthermore, this entry was also added to the Tudor domain clan (CL0049).”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingnone given

    Pfam clan CL0063 encompasses 198 entries belonging to the FAD/NAD(P)-binding Rossmann fold superfamily.

    “For example, Pfam clan CL0063 encompasses 198 entries belonging to the FAD/NAD(P)-binding Rossmann fold superfamily.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingcitation

    Pfam sequence and residue coverage of UniProtKB has remained fairly constant since Pfam 29.0, with sequence coverage 76.1% and residue coverage 54.8% in Pfam 29.0.

    “The Pfam sequence and residue coverage of UniProtKB has remained fairly constant since Pfam 29.0 (released in 2015, wherein the corresponding UniProtKB sequence coverage was 76.1% and residue coverage was 54.8%) (2).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [2] Finn, 2016, The Pfam protein families database: towards a more sustainable future, Nucleic Acids Res.

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Findingreasoning

    ECOD was used to update DUF5328 to Cas6b_C (Pfam: PF17262) and was also used to rename DUF2945 to Hva1_TUDOR (Pfam: PF11160), after noting that the hypervirulence-associated protein 1 contains this domain.

    “As previously mentioned, ECOD was used to update DUF5328 to Cas6b_C (Pfam: PF17262) and was also used to rename DUF2945 to Hva1_TUDOR (Pfam: PF11160), after noting that the hypervirulence-associated protein 1 contains this domain.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Findingnone given

    The Peptidase_M14 entry (Pfam: PF00246) contains both active and inactive homologues and can be subdivided into at least four subfamilies.

    “For example, the Peptidase_M14 entry (Pfam: PF00246) contains both active and inactive homologues and can be subdivided into at least four subfamilies.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingreasoning

    Detailed comparisons between Pfam (version 31.0) and a subset of ECOD (version 29) have guided the inclusion of the majority of the new Pfam entries, improved family definitions and entries added to clans.

    “Detailed comparisons between Pfam (version 31.0) and a subset of ECOD (version 29) have guided the inclusion of the majority of the new Pfam entries, improved family definitions and entries added to clans.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The F-group level in ECOD is primarily derived from Pfam, based on sequence similarity.

  • May combine two claims.

    Findingcitation

    Based on the ECOD domain boundaries for PDB accession 4z7k (ECOD domain: e4z7kB2), the N-terminal region of DUF5328 was removed in line with the structural domain; the modified domain corresponded to the C-terminal domain of Cas6b proteins and was renamed to Cas6b_C (Pfam: PF17262).

    “Based on the ECOD domain boundaries for the Protein Data Bank (PDB) (15,16) accession 4z7k (ECOD domain: e4z7kB2), we removed the N-terminal region of DUF5328 in line with the structural domain. This modified domain corresponded to the C-terminal domain of Cas6b proteins and was accordingly renamed to Cas6b_C (Pfam: PF17262), adding both functional annotation and literature references.”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [15] Berman, 2003, Announcing the worldwide Protein Data Bank, Nat. Struct. Biol.

      Not checked in the free run.

    • cites [16] Berman, 2007, The worldwide Protein Data Bank (wwPDB): ensuring a single, uniform archive of PDB data, Nucleic Acids Res.

      Not checked in the free run.

  • May combine two claims.

    Findingcitation

    Pfam 32.0 now covers 87.1% of sequences and 73.8% of all residues represented by known structures found in PDB (as of 9 May 2018).

    “As such, Pfam 32.0 now covers 87.1% of sequences and 73.8% of all residues represented by known structures found in PDB (15,16) (9 May 2018).”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [15] Berman, 2003, Announcing the worldwide Protein Data Bank, Nat. Struct. Biol.

      Not checked in the free run.

    • cites [16] Berman, 2007, The worldwide Protein Data Bank (wwPDB): ensuring a single, uniform archive of PDB data, Nucleic Acids Res.

      Not checked in the free run.

  • May combine two claims.

    Findingown data

    Although entries belonging to clans represent less than 40% of entries in Pfam, they correspond to 74% of all sequence regions annotated by Pfam, an increase of 6% since Pfam 29.0.

    “Although this number represents less than 40% of entries in Pfam, it corresponds to 74% of all sequence regions annotated by Pfam (an increase of 6% since Pfam 29.0).”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Findingnone given

    Hierarchical classifications of proteins such as Pfam clans have expanded substantially over the past 5 years, with 74% of all Pfam regions now belonging to a clan.

    “In the meantime, hierarchical classifications of proteins such as Pfam clans, have expanded substantially over the past 5 years, with 74% of all Pfam regions now belonging to a clan.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingreasoning

    HEAT repeats, including the Importin HEAT-like repeat, have six different and specific entries reflecting their high sequence diversity.

    “This can be illustrated by the HEAT repeats, including the Importin HEAT-like repeat, which have six different and specific entries reflecting their high sequence diversity.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Sequence and residue coverage of the reference proteomes in Pfam 32.0 is 74.5% (an increase of 1.0%) and 50.1% (an increase of 3.1%), respectively.

    “The sequence and residue coverage of the reference proteomes in Pfam 32.0 is 74.5% (an increase of 1.0%) and 50.1% (an increase of 3.1%), respectively.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingnone given

    There are 628 clans in total within Pfam 32.0, with 74 new clans added since Pfam 29.0.

    “There are 628 clans in total within Pfam 32.0, with 74 new clans added since Pfam 29.0.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingreasoning

    The coverage statistics have been maintained as a result of new sequences being matched by existing Pfam entries, and the generation of 1,664 new entries.

    “Thus, these coverage statistics have been maintained as a result of new sequences being matched by existing Pfam entries, and the generation of 1,664 new entries.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data

    Of the 313 unique Pfam authors, just over half (161) have an associated ORCID.

    “Of the 313 unique Pfam authors, just over half (161) have an associated ORCID.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Findingown data

    Types family and domain are the most common (6,248 and 11,177 entries, respectively), comprising over 97.2% of all entries.

    “Types family and domain are the most common (6,248 and 11,177 entries, respectively), comprising over 97.2% of all entries.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingreasoning

    Cas6b_N adopts a similar structure to Cas6b_C and was therefore also added to clan CL0362.

    “Cas6b_N adopts a similar structure to Cas6b_C and was therefore also added to clan CL0362.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

    Interpretationreasoning

    Linking DUFs to functionally characterized members within a clan can provide important insights into the potential role of these functionally uncharacterized sequences.

    “Linking DUFs to functionally characterized members within a clan can provide important insights into the potential role of these functionally uncharacterized sequences.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationcitation

    These new sequences cover an ever broader range of taxonomies, with the UniProtKB redundancy procedures ensuring that growth reflects increased diversity, rather than additional strains of the same bacterial species.

    “These new sequences cover an ever broader range of taxonomies, with the UniProtKB redundancy procedures (9) ensuring that growth reflects increased diversity, rather than, for instance, additional strains of the same bacterial species.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [9] Bursteinas, 2016, Minimizing proteome redundancy in the UniProt Knowledgebase, Database (Oxford)

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    Carboxypeptidase A1 favours substrate residues in the cleavage site with aromatic or branched side chains, while carboxypeptidase E favours basic amino acids.

    “Of these, carboxypeptidase A1 favours substrate residues in the cleavage site with aromatic or branched side chains, while carboxypeptidase E favours basic amino acids (7).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [7] Rawlings, 2018, The MEROPS database of proteolytic enzymes, their substrates and inhibitors in 2017 and a comparison with peptidases in the PANTHER database, Nucleic Acids Res.

      Not checked in the free run.

  • May combine two claims.

    Interpretationreasoning

    The addition of hundreds of new entries has made little impact on the fraction covered, primarily due to the expansion of the sequence databases, thereby demonstrating the continual need for curation of new Pfam entries.

    “The addition of hundreds of new entries has made little impact on the fraction covered, primarily due to the expansion of the sequence databases, thereby demonstrating the continual need for curation of new Pfam entries.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning

    The tendency of repeated sequences to diverge is especially true for flanking units, meaning that even long models representing the entire TR often fail to completely represent the TR region in Pfam.

    “Indeed, the tendency of repeated sequences to diverge is especially true for flanking units, meaning that even long models representing the entire TR often fail to completely represent the TR region in Pfam.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The tendency of repeated sequences to diverge is especially true for flanking units.

  • May combine two claims.

    Interpretationreasoning

    No single approach ensures the proper identification of the correct periodicity of the repeat nor the representation of the TR in its entirety.

    “No single approach ensures the proper identification of the correct periodicity of the repeat nor the representation of the TR in its entirety.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Naturally occurring proteins are characterised by low sequence conservation between repeating units.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    Perfect repeats in engineered sequences exist and often correlate with a tendency to be unstructured.

    “Nevertheless, perfect repeats in engineered sequences do exist and ironically, often correlate with a tendency to be unstructured (18).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [18] Jorda, 2010, Protein tandem repeats - the more perfect, the less structured, FEBS J.

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    Pfam is part of the InterPro consortium, an amalgamation of 14 different expert databases.

    “Pfam is part of the InterPro consortium, an amalgamation of 14 different expert databases (20), which present a single, unified view of regions of functional importance in protein sequences.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [20] Finn, 2017, InterPro in 2017-beyond protein family and domain annotations, Nucleic Acids Res.

      Not checked in the free run.

26 of 26

  • The paper relies on this without support.

    The F-group level in ECOD is primarily derived from Pfam, based on sequence similarity.

    “The F-group level (family level) in ECOD is primarily derived from Pfam, i.e. based on sequence similarity.”

  • The paper relies on this without support.

    Naturally occurring proteins are characterised by low sequence conservation between repeating units.

    “On the other hand, naturally occurring proteins are characterised by low sequence conservation between repeating units.”

  • The paper relies on this without support.

    The tendency of repeated sequences to diverge is especially true for flanking units.

    “Indeed, the tendency of repeated sequences to diverge is especially true for flanking units, meaning that even long models representing the entire TR often fail to completely represent the TR region in Pfam.”

  • The paper relies on this without support.

    The majority of annotation updates have come from InterPro curators.

    “The majority of annotation updates have nonetheless come from InterPro curators.”
4 of 4

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • Pfam annotations are now provided as an additional results set to aid comparison with other resources that use Pfam for annotation, such as IMG/M.

    “While previously available as part of the InterPro results, Pfam (39) annotations are now provided as an additional results set to aid comparison with other resources that use Pfam for annotation, such as IMG/M.”

    From MGnify: the microbiome analysis resource in 2020 (2019) · cites this paper as [39]

Cites

Flagged first, then by citations

Research works printed but not linked: 1

  1. Sonnhammer, 1997, Pfam: a comprehensive database of protein domain families based on seed alignments, Proteins, 28, 405

Cited by

Sorted by citations

Related

Authors
El-Gebali, S., Mistry, J., Bateman, A., Eddy, S. R., Luciani, A., Potter, S. C., Qureshi, M., Richardson, L. J., Salazar, G. A., Smart, A., Sonnhammer, E. L. L., Hirsh, L., Paladin, L., Piovesan, D., Tosatto, S. C. E., & Finn, R. D.
Affiliations
  • European Bioinformatics Institute (GB)
  • Howard Hughes Medical Institute (US)
  • Stockholm University (SE)
  • University of Padua (IT)
Corresponding author
Finn, Robert D
ORCID
Venue
Nucleic Acids Research
Publisher
Oxford University Press (OUP)
Published
2018-10-24
Volume
47
Issue
D1
Pages
D427-D432
ISSN
  • 0305-1048 (ISSN-L)
  • 1362-4962
DOI
10.1093/nar/gky995
PMC
PMC6324024
OpenAlex
W2898402099
Semantic Scholar
7f19972754ac0c15329666b3a6efbf569b27d8d5
Cited by
4,602
References
25
Citations by year
2026: 162 · 2025: 253 · 2024: 347 · 2023: 531 · 2022: 749 · 2021: 1161 · 2020: 1030 · 2019: 355 · 2018: 8 · 2017: 1 · 2016: 1
FWCI
236.49
Type
journal article
Language
en
Open access
Licence
Locations
  • Nucleic Acids Research (publishedVersion, cc-by)
  • Europe PMC (PubMed Central) (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • Padua Research Archive (University of Padova) (submittedVersion, cc-by)
  • Greater South Information System
Funders
  • Horizon 2020 (654039)
  • Wellcome Trust (108433/Z/15/Z)
  • Biotechnology and Biosciences Research Council (BB/L024136/1)
  • European Cooperation in Science and Technology (BM1405)
  • National Institutes of Health (R01-HG009116)
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Genomics and Phylogenetic Studies
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Protein Structure and Dynamics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Machine Learning in Bioinformatics
Keywords
Pfam, protein families, protein domains, domain boundaries, functional annotation, tandem repeats, ECOD, Sequence Ontology, ORCID, database curation
MeSH terms
Protein Domains, Proteins, Repetitive Sequences, Amino Acid, Databases, Protein, Molecular Sequence Annotation

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.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-10.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-10.

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