1. Marie-Claude.Blatter@isb-sib.ch Swiss-Prot group, Geneva SIB Swiss Institute of Bioinformatics Protein sequence databases: dissemination of protein knowledge http://education.expasy.org/cours/UniProt/

2. Menu Introduction Nucleic acid sequence databases ENA, GenBank, DDBJ Protein sequence databases UniProt databases (UniProtKB) NCBI protein databases (NCBInr, RefSeq…)

3. Protein sequences are the fundamental determinants of biological structure and function. http://www.ncbi.nlm.nih.gov/protein

4. Challenge Flood of data -> need to be stored, curated and made available for analysis and knowledge discovery

5. TrEMBL Genpept Swiss-Prot RefSeq PRF Ensembl CCDS UniParc UniProtKB PDB(PIR) (IPI) UniMES TPA Challenge (1) Many different protein sequence databases NCBInr

7. These identifiers are all pointing to the same TP53 protein sequence (p53) ! P04637, NP_000537, ENSG00000141510, CCDS11118, UPI000002ED67, IPI00025087, HIT000320921, XP_001172091, DD954676, JT0436, etc. Challenge (1bis) Different protein sequence databases : many identifiers for the same protein sequence

8. A HUPO test sample study reveals common problems in mass spectrometry–based proteomics PubMed 19448641 (2009)19448641 A single mass spectrometry experiment can identified up to about 4000 proteins (15’000 peptides) Protein databases vary greatly in terms of their curation, completeness and comprehensiveness (search with different protein databases = could get different results). Only 7 labs (on 27) were able to identify the 20 human proteins present in a sample, mainly due to the fact that the search engines used cannot distinguish among different identifiers for the same protein…

9. Nucleic Acids Res. 2010 ; 38(Database issue): D633–D639. ‘Examining links from the perspective of PubMed, we found that only a small fraction of published articles are linked to human genes (Entrez Gene).’ Challenge (3) (protein) sequence annotation

10. Journals do not (SHOULD NOT) accept a paper dealing with a nucleic acid sequence if the ENA/GenBank/DDBJ AC number is not available… ‘journal publishers generally require deposition prior to publication so that an accession number can be included in the paper.’ http://www.ncbi.nlm.nih.gov/books/bv.fcgi?highlight=refseq&rid=handbook.section.GenBank_ASM#GenBank_ASM.RefSeq …not the case for protein sequences !!! no more the case for a lot of genomes !!!

11. Protein sequence origin…

12. More than 99 % of the protein sequences are derived from the translation of nucleotide sequences (genomes and/or cDNAs) sequencing quality coding sequence (CDS) annotation accuracy gene prediction quality

14. … ~ 2500 genomes sequenced (single organism, varying sizes, including virus) … ~ 5’000 ongoing genome sequencing projects

15. http://www.ncbi.nlm.nih.gov/genomes/static/gpstat.html http://www.ncbi.nlm.nih.gov/genomes/GenomesHome.cgi?taxid=10239&hopt=stat ~ 50-100 genomes/month + ~2’500 viral genomes => Total ~ 5’000 genomes

16. … ~ 2500 genomes sequenced (single organism, varying sizes, including virus) … ~ 5’000 ongoing genome sequencing projects … cDNAs sequencing projects (ESTs or cDNAs) … metagenome sequencing projects = environmental samples: multiple ‘unknown’ organisms,

17. Metagenomics Metagenomics study of genetic material recovered directly from environmental samples Global Ocean Sampling (C. Venter) 1ml sea water: 1 mo bacteria and 10 mo virus Whale fall (AAFZ00000000.1) Soil, sand beach, New-York air, … Human fluids, mouse gut (millions of bacteria within human body) Water treatment industry… Lists of projects: http://www.ncbi.nlm.nih.gov/genomes/lenvs.cgi Venter’s Sorcerer II

18. … ~ 2500 genomes sequenced (single organism, varying sizes) … ~ 5’000 ongoing genome sequencing projects … cDNAs sequencing projects (ESTs or cDNAs) … metagenome sequencing projects … personal human genomes new generation sequencers : Illumina: 25 billions of bp /day;

19. http://www.youtube.com/watch?v=mVZI7NBgcWM …2700 genomes in 2010, 30’000 genomes in 2011 ? 2’000’000 $ (2007) 70’000’000 $ (diploid, 2007) 3’000’000’000 $ (public consortium, 2000) 300’000’000 $ (Celera, 2000) 2010

20. But…we known now that his apoE allele is the one associated with increased risk for Alzheimer and that he has the ‘blue eye’ allele…

21. apoE gene (Ensembl genome browser)

22. New projects (homo sapiens) 1000 genomes (first publication, October 2010) Multiple personal genomes (sexual cells, lymphoid cells, cancer cells…) International cancer genome consortium (www.icgc.org).www.icgc.org They look at the most common cancers and for each they sequence the genome of 500 patients with cancer and 500 healthy individuals…. How to define the human proteome ??? Which sequences ???

23. How many proteins-coding genes at the end?

24. 190‘500'025'042 1st estimate: ~30 million species (1.8 million named) 2 nd estimate: 20 million bacteria/archea x 4'000 genes 1 million protists x 6'000 genes 5 million insects x 14'000 genes 2 million fungi x 6'000 genes 0.5 million plants x 20'000 genes 0.5 million molluscs, worms, arachnids, etc. x 20'000 genes 0.1 million vertebrates x 25'000 genes The calculation: 2x10 7 x4000+1x10 6 x6000+5x10 6 x14000+2x10 6 x6000+5x10 5 x20000+5x10 5 x20000+1x10 5 x25000 +20000 (Craig Venter)+ 42(Douglas Adam) + …

25. About 190 billions of proteins (?) About 14.0 millions of ‘known’ protein sequences in 2011 (from ~300’000 species) More than 99 % of the protein sequences are derived from the translation of nucleotide sequences Less than 1 % direct protein sequencing (Edman, MS/MS…) -> It is important that protein database users know where the protein sequence comes from…

26. cDNAs, ESTs, genes, genomes, … Nucleic acid sequence databases The ideal life of a sequence … Protein sequence databases

27. Menu Introduction Nucleic acid sequence databases ENA/GenBank, DDBJ Protein sequence databases UniProt databases (UniProtKB) NCBI protein databases

28. ENA (EMBL-Bank) GenBank DDBJ DNA Data Bank of Japan archive of primary sequence data and corresponding annotation submitted by the laboratories that did the sequencing. European Nucleotide Archive

29. http://www.insdc.org/ ENA/GenBank/DDBJ

30. Serve as archives : ‘nothing goes out’ Contain all public sequences derived from: – Genome projects (> 80 % of entries) – Sequencing centers (cDNAs, ESTs…) – Individual scientists ( 15 % of entries) – Patent offices (i.e. European Patent Office, EPO) Currently: ~210x10 6 sequences, ~320 x10 9 bp; Sequences from > 300’000 different species; ENA/GenBank/DDBJ

31. Archival databases: -Can be very redundant for some loci -Sequence records are owned by the original submitter and can not be alterered by a third party (except TPA)

32. taxonomy Cross-references references accession number

33. CDS annotation (Prediction or experimentally determined) sequence CDS CoDing Sequence (proposed by submitters)

34. The hectic life of a sequence … cDNAs, ESTs, genes, genomes, … ENA, GenBank, DDBJ Data not submitted to public databases, delayed or cancelled… with or without annotated CDS provided by authors CDS CoDing Sequence portion of DNA/RNA translated into protein (from Met to STOP) Experimentally proved or derived from gene prediction !!! not so well documented !!!

35. CONTIG --------------------------------------------------------------------------------------CGANGGCCTATCAACA ATG AAAGGTCGAAACCTG Genomic AGCTACAAACAGATCCTTGATAATTGTCGTTGATTTTACTTTATCCTAAATTTATCTCAAAAATGTTGAAATTCAGATTCGTCAAGCGAGGGCCTATCAACA ATG -AAGGTCGAAACCTG *** ************ ** * ************** CONTIG CGTTTACTCCGGATACAAGATCCACCCAGGACACGGNAAAGAGACTTGTCCGTACTGACGGAAAG------------------------------------------------------- Genomic CGTTTACTCCGGATACAAGATCCACCCAGGACACGG-AAAGAGACTTGTCCGTACTGACGGAAAGGTGAGTTCAGTTTCTCTTTGAAAGGCGTTAGCATGCTGTTAGAGCTCGTAAGGTA ************************************ **************************** CONTIG ------------------------------------------------------------------------------------------------------------------------ Genomic TATTGTAATTTTACGAGTGTTGAAGTATTGCAAAAGTAAAGCATAATCACCTTATGTATGTGTTGGTGCTATATCTTCTAGTTTTTAGAAGTTATACCATCGTTAAGCATGCCACGTGTT CONTIG ----------------------------------------------GTCCAAATCTTCCTCAGTGGAAAGGCACTCAAGGGAGCCAAGCTTCGCCGTAACCCACGTGACATCAGATGGAC Genomic GAGTGCGACAAACTACCGTTTCATGATTTATTTATTCAAATTTCAGGTCCAAATCTTCCTCAGTGGAAAGGCACTCAAGGGAGCCAAGCTTCGCCGTAACCCACGTGACATCAGATGGAC ************************************************************************** CONTIG TGTCCTCTACAGAATCAAGAACAAGAAG---------------------------------------------GGAACCCACGGACAAGAGCAAGTCACCAGAAAGAAGACCAAGAAGTC Genomic TGTCCTCTACAGAATCAAGAACAAGAAGGTACTTGAGATCCTTAAACGCAGTTGAAAATTGGTAATTTTACAGGGAACCCACGGACAAGAGCAAGTCACCAGAAAGAAGACCAAGAAGTC **************************** *********************************************** CONTIG CGTCCAGGTTGTTAACCGCGCCGTCGCTGGACTTTCCCTTGATGCTATCCTTGCCAAGAGAAACCAGACCGAAGACTTCCGTCGCCAACAGCGTGAACAAGCCGCTAAGATCGCCAAGGA Genomic CGTCCAGGTTGTTAACCGCGCCGTCGCTGGACTTTCCCTTGATGCTATCCTTGCCAAGAGAAACCAGACCGAAGACTTCCGTCGCCAACAGCGTGAACAAGCCGCTAAGATCGCCAAGGA ************************************************************************************************************************ CONTIG TGCCAACAAGGCTGTCCGTGCCGCCAAGGCTGCTNCCAACAAG----------------------------------------------------------------------------- Genomic TGCCAACAAGGCTGTCCGTGCCGCCAAGGCTGCTGCCAACAAGGTAAACTTTCTACAATATTTATTATAAACTTTAGCATGCTGTTAGAGCTTGTAAGGTATATGTGATTTTACGAGTGT ********************************** ******** CONTIG -------------------------------------------------------------------------------------------------------------------GNAAA Genomic GTTATTTGAAGCTGTAATATCAATAAGCATGTCTCGTGTGAAGTCCGACAATTTACCATATGCATGAAATTTAAAAACAAGTTAATTTTGTCAATTCTTTATCATTGGTTTTCAGGAAAA * *** CONTIG GAAGGCCTCTCAGCCAAAGACCCAGCAAAAGACCGCCAAGAATNTNAAGACTGCTGCTCCNCGTGTCGGNGGAAANCGA TAA ACGTTCTCGGNCCCGTTATTGTAATAAATTTTGTTGAC Genomic GAAGGCCTCTCAGCCAAAGACCCAGCAAAAGACCGCCAAGAATGTGAAGACTGCTGCTCCACGTGTCGGAGGAAAGCGA TAA ACGTTCTCGGTCCCGTTATTGTAATAAATTTTGTTGAC ******************************************* * ************** ******** ***** **** * *********** *************************** CONTIG C----------------------------------------------------------------------------------------------------------------------- Genomic CGTTAAAGTTTTAATGCAAGACATCCAACAAGAAAAGTATTCTCAAATTATTATTTTAACAGAACTATCCGAATCTGTTCATTTGAGTTTGTTTAGAATGAGGACTCTTCGAATAGCCCA * CoDing Sequence Alignment between a mRNA and a genomic sequence exon intron

36. CDS translation provided by ENA CDS provided by the submitters The first Met !

37. UCSC: human EPO 5’ 3’ mRNAs and their corresponding CDS annotation (from EMBL/GenBank/DDBJ) contig

38. mRNAs and their corresponding CDS annotation (from EMBL/GenBank/DDBJ)

39. Very rarely done…

40. Complete genome (submitted) but only ~ 2,000 CDS/proteins available !

41. http://www.ebi.ac.uk/swissprot/sptr_stats/index.html …annotated CDS in UniProtKB

42. From nucleic acid to amino acid sequences databases….

43. The hectic life of a protein sequence … cDNAs, ESTs, genomes, … ENA, GenBank, DDBJ Data not submitted to public databases, delayed or cancelled… …if the submitters provide an annotated Coding Sequence (CDS) (1/10 ENA entries) Protein sequence databases Nucleic acid databases Gene prediction RefSeq, Ensembl no CDS RefSeq, Ensembl and other

44. Why doing things in a simple way, when you can do it in a very complex one ?

45. The hectic life of a sequence … TrEMBL Genpept CoDing Sequences provided by submitters cDNAs, ESTs, genomes, … ENA, GenBank, DDBJ Data not submitted to public databases, delayed or cancelled… Swiss-Prot RefSeq PRF Scientific publications derived sequences Ensembl CCDS UniParc UniProtKB PDB(PIR) + all ‘species’ specific databases (EcoGene, TAIR, …) (IPI) UniMES CoDing Sequences provided by submitters and gene prediction TPA

46. Major ‘general’ protein sequence database ‘sources’ UniProtKB: Swiss-Prot + TrEMBL NCBI-nr: Swiss-Prot + TrEMBL + GenPept + PIR + PDB + PRF + RefSeq + TPA PIR PDBPRF UniProtKB/Swiss-Prot: manually annotated protein sequences (12’000 species) UniProtKB/TrEMBL: submitted CDS (ENA) + automated annotation; non redundant with Swiss-Prot (300’000 species) GenPept: submitted CDS (GenBank); redundant with Swiss-Prot (300’000 species ?) PIR: Protein Information Ressource; archive since 2003; integrated into UniProtKB PDB: Protein Databank: 3D data and associated sequences PRF: Protein Research Foundation journal scan of ‘published’ peptide sequences RefSeq: Reference Sequence for DNA, RNA, protein + gene prediction + some manual annotation (11’000 species) TPA: Third part annotation Integrated resources ‘cross-references’ Resources kept separated TPA not complete !!! (only entries created before 2007 ?)

47. Swiss-Prot TrEMBL Look for EPO (homo sapiens) www.uniprot.org

48. Look for EPO (homo sapiens) Swiss-Prot GenPept RefSeq GenPept

49. Menu Introduction Nucleic acid sequence databases ENA-Bank/GenBank, DDBJ Protein sequence databases UniProt databases (UniProtKB) NCBI protein databases

50. UniProt UniProt consortium EBI : European Bioinformatics Institute (UK) SIB : Swiss Institute of Bioinformatics (CH) PIR : Protein information resource (US)

51. UniProt databases

52. UniProtKB UniProtKB: protein sequence knowledgebase, 2 sections UniProtKB/Swiss-Prot and UniProtKB/TrEMBL (query, Blast, download) (~14 mo entries) UniParc UniParc: protein sequence archive (ENA equivalent at the protein level). Each entry contains a protein sequence with cross- links to other databases where you find the sequence (active or not). Not annotated (query, Blast, download) (~25mo entries) UniRef UniRef: 3 clusters of protein sequences with 100, 90 and 50 % identity; useful to speed up sequence similarity search (BLAST) (query, Blast, download) (UniRef100 10 mo entries; UniRef90 7 mo entries; UniRef50 3.3 mo entries) UniMES UniMES: protein sequences derived from metagenomic projects (mostly Global Ocean Sampling (GOS)) (download) (8 mo entries, included in UniParc)

53. UniProtKB an encyclopedia on proteins composed of 2 sections UniProtKB/TrEMBL and UniProtKB/Swiss-Prot unreviewed and reviewed automatically annotated and manually annotated released every 4 weeks

54. UniProtKB from EMBL to TrEMBL UniProtKB protein sequence data are mainly derived from EMBL (CDS) but also from Ensembl, RefSeq, model organism databases (MODs; e.g. TAIR) and PDB. Data from the PIR database have been integrated in UniProtKB since 2003.

55. UniProtKB/Swiss-Prot and UniProtKB/TrEMBL give access to all the protein sequences which are available to the public. However, UniProtKB excludes the following protein sequences: - Most non-germline immunoglobulins and T-cell receptors - Synthetic sequences - Most patent application sequences - Small fragments encoded from nucleotide sequence (<8 amino acids) - Pseudogenes* - Fusion/truncated proteins - Not real proteins * many putative pseudogene sequences (which are tagged as potential pseudogenes) may be expected to remain in UniProtKB for some time as it can be difficult to prove the non-existence of a protein

56. Data increase in UniProtKB

57. TrEMBL EMBL Automated extraction of protein sequence (translated CDS), gene name and references.+ Automated annotation

58. One protein sequence One species Automated annotation Keywords and Gene Ontology Automated annotation Function, Subcellular location, Catalytic activity, Sequence similarities… Automated annotation transmembrane domains, signal peptide… Cross-references to over 125 databases References Protein and gene names Taxonomic information UniProtKB/TrEMBL www.uniprot.org

59. UniProtKB: from EMBL to TrEMBL Automated annotation 1. Protein sequence 2. Biological information

60. Protein sequence - The quality of UniProtKB/TrEMBL protein sequences is dependent on the information provided by the submitter of the original nucleotide entry (CDS). - 100% identical sequences (same lenght, same organism are merged automatically). Biological information Sources of annotation -Provided by the submitter (EMBL, PDB, TAIR…) -From automated annotation (SAAS: automated generated annotation rules) -From automated annotation (UniRule; manually generated annotation rules) UniProtKB/TrEMBL

62. Protein sequence - The quality of UniProtKB/TrEMBL protein sequences is dependent on the information provided by the submitter of the original nucleotide entry (CDS). - 100% identical sequences (same length, same organism are merged automatically). Biological information Sources of annotation -Provided by the submitter (EMBL, PDB, TAIR) -From automated annotation (SAAS: automated generated annotation rules) -From automated annotation (UniRule; manually generated annotation rules) UniProtKB/TrEMBL

63. Automatic annotation in UniProtKB/TrEMBL System Rule creation TriggerAnnotationsScope SAASautomaticInterProcomments, KWall taxa UniRulesmanualInterPro* protein names, comments, features, KW, GO terms all taxa * Flexibility to create custom signatures for InterPro as required

64. SAAS Rules are derived from the UniProtKB/Swiss-Prot manual annotation. Fully automated rule generation based on C4.5 decision tree algorithm. One annotation, one rule. Precision calculated for each rule vs UniProtKB/Swiss-Prot. High stringency – require 99% or greater estimated precision on UniProtKB/TrEMBL to generate annotation. Rules are produced, updated and validated at each release. UniProtKB/TrEMBL

67. UniRules (RuleBase, HAMAP, PIRSF) Rules of varying complexity: annotation varies from simple KW attribution to complete annotation as for UniProtKB/Swiss-Prot Rules are manually curated:  From SAAS rules as input  From UniProtKB/Swiss-Prot annotation and InterPro match data, taxonomy information – continuously reported to curators  From literature based curation of characterized families - with the possibility to create new signatures for specific functional groups Rules are continuously monitored – validation on UniProtKB/Swiss-Prot – 97% confidence HAMAP is also used for the annotation of some UniProtKB/Swiss-Prot entries Conditions Annotations

68. UniRule – HAMAP

70. SAAS – automatically generated annotation rules for comments, KWs - Tested on UniProtKB/Swiss-Prot UniRule – manually curated annotation rules (e.g. HAMAP) –annotation varies from simple KWs to full annotation –start point can be SAAS rules, InterPro reports, literature-based curation of protein families –possibility to create custom signatures -> InterPro Automatic annotation of UniProtKB/TrEMBL is refreshed, and validated, each UniProtKB release – validation using UniProtKB/Swiss-Prot as reference. ~10% of the rules are ‘refreshed’ at each release. The source of each annotation is indicated - users can access rule logic Automatic annotation in UniProtKB/TrEMBL - Summary

71. Current status – coverage of UniProtKB/TrEMBL SystemRulesCoverage* SAAS168417.2% UniRules RuleBase110823.0% PIR name / site rules 1420.26% HAMAP10874.7% * Proportion of entries with at least one annotation from the specified system UniProtKB/TrEMBL 2010_12: 12,769,092 entries, all systems combined, 33%

72. Current status - coverage of UniProtKB/TrEMBL UniProt release 2010_10 included annotations from 7767 SAAS rules, 1814 UniRules

73. GO annotation - KW2GO -InterPro2GO -HAMAP2GO

74. UniProtKB from TrEMBL to Swiss-Prot Once manually annotated and integrated into Swiss- Prot, the entry is deleted from TrEMBL -> minimal redundancy

75. TrEMBL EMBL Automated extraction of protein sequence (translated CDS), gene name and references.+ Automated annotation Manual annotation of the sequence and associated biological information Swiss-Prot

76. MSKEKFERTKPHVNVGTIGHVDHGKTTLTAAITTVLAKTYGGAAR AFDQIDNAPEEKARGITINTSHVEYDTPTRHYAHVDCPGHADYVK NMITGAAQMDGAILVVAATDGPMPQTREHILLGRQVGVPYIIVFL NKCDMVDDEELLELVEMEVRELLSQYDFPGDDTPIVRGSALKALE GDAEWEAKILELAGFLDSYIPEPERAIDKPFLLPIEDVFSISGRG TVVTGRVERGIIKVGEEVEIVGIKETQKSTCTGVEMFRKLLDEGR AGENVGVLLRGIKREEIERGQVLAKPGTIKPHTKFESEVYILSKD EGGRHTPFFKGYRPQFYFRTTDVTGTIELPEGVEMVMPGDNIKMV VTLIHPIAMDDGLRFAIREGGRTVGAGVVAKVLG One protein sequence One gene One species Manual annotation Keywords and Gene Ontology Manual annotation Function, Subcellular location, Catalytic activity, Disease, Tissue specificty, Pathway… Manual annotation Post-translational modifications, variants, transmembrane domains, signal peptide… Cross-references to over 125 databases References Protein and gene names Taxonomic information Alternative products: protein sequences produced by alternative splicing, alternative promoter usage, alternative initiation… UniProtKB/Swiss-Prot www.uniprot.org

77. UniProtKB: from TrEMBL to Swiss-Prot Manual annotation 1. Protein sequence (merge available CDS, annotate sequence discrepancies, report sequencing mistakes…) 2. Biological information (sequence analysis, extract literature information, ortholog data propagation, …)

78. UniProtKB 1- Sequence curation

79. The displayed protein sequence …canonical, representative, consensus…

80. The displayed sequence is the most prevalent protein sequence and/or the protein sequence which is also found in orthologous species. The displayed sequence is generally derived from the translation of the genomic sequence (when available). Sequence differences are documented. 1 entry 1 gene (1 species) 1 displayed sequence (annotation of alternative sequences, when available) UniProtKB/Swiss-Prot protein sequence annotation ‘Merging/Redundancy policy’: a gene-centric view of protein space

81. What is the current status? At least 20% of Swiss-Prot entries required a minimal amount of curation effort so as to obtain the “correct” sequence. Typical problems –unsolved conflicts; –uncorrected initiation sites; –frameshifts; –other ‘problems’

83. … once a gene on chromosome 11…

84. Quality of protein information from genome projects Lets look at proteins originating from genome projects: –Drosophila: the paradigm of a curated genome should look like (thanks to FlyBase) : only 1.8% of the gene models conflict with Swiss-Prot sequences; –Arabidopsis: a typical example of a genome where a lot of annotation was done when it was sequenced, but no update since then (at least in the public view): 20% of the gene models are erroneous; –Tetraodon nigroviridis: the typical example of a quick and dirty automatic run through a genome with no manual intervention: >90% of the gene models produce incorrect proteins. –Bacteria and Archaea have almost no splicing, so predictions are “easier”, however errors are still made… Start codons, missed small proteins (<100aa)…

85. UniProtKB/Swiss-Prot Protein sequence annotation

86. Example of problem (derived from gene prediction pipeline) Ensembl completes the human ‘proteome’ by predicting/annotating missing genes according to orthologous sequences.. ID URAD_HUMAN Unreviewed; 171 AA. AC A6NGE7; DT 24-JUL-2007, integrated into UniProtKB/TrEMBL. DT 24-JUL-2007, sequence version 1. DT 02-OCT-2007, entry version 3. DE 2-oxo-4-hydroxy-4-carboxy-5-ureidoimidazoline decarboxylase homolog DE (OHCU decarboxylase homolog) (Parahox neighbour). GN Name=PRHOXNB; … DR EMBL; AL591024; -; NOT_ANNOTATED_CDS; Genomic_DNA. DR Ensembl; ENSG00000183463; Homo sapiens. DR HGNC; HGNC:17785; PRHOXNB. PE 4: Predicted; In primates the genes coding for the enzymes for the degradation of uric acid were inactivated and converted to pseudogenes.

87. Producing a clean set of sequences is not a trivial task; It is not getting easier as more and more types of sequence data are submitted;

88. The ‘Protein existence’ tag indicates what is the evidence for the existence of a given protein; Different qualifiers: 1. Evidence at protein level (~18%) (MS, western blot (tissue specificity), immuno (subcellular location),…) 2. Evidence at transcript level (~19%) 3. Inferred from homology (~58 %) 4. Predicted (~5%) 5. Uncertain (mainly in TrEMBL) ‘Protein existence’ tag http://www.uniprot.org/docs/pe_criteria

90. In order to avoid ‘pseudogenes’ and most of the unprobable protein sequences, you can filter your query and avoid sequences with ‘protein existence tag’ = ‘Uncertain’

92. The ‘alternative’ sequence(s)

93. UniProtKB/Swiss-Prot 1 entry 1 gene (1 species) Annotation of the sequence differences (including conflicts, polymorphisms, splice variants etc..) -> annotation of protein diversity

94. Multiple alignment of the end of the available GCR sequences Annotation of the sequence differences (protein diversity) 1 entry 1 gene (1 species) …and natural variants

95. P04150 www.uniprot.org

96. UniProtKB (and RefSeq) do under-represent alternatively spliced products According to PMID:21307931: alternative splicing seems to occur at more than 90% of protein-coding genes (might not always modify the protein sequence). Transcript variants are only made when there is information available on the full-lenght nature of the product; if multiple, alternate exons are found through the lenght of the gene, no assumption is made about the combination of the alternate exons that may exist in vivo. Uncertain alternative sequences (confirmed by only one cDNA) are tagged with ‘No experimental confirmation available’ http://www.ncbi.nlm.nih.gov/books/NBK50679/#RefSeqFAQ.what_does_a_reviewed_status_me

97. Available in separated files! Important remark > 30’000 additional sequences (total)

98. The ‘alternative’ sequence(s) not ‘directly available’ for a lot of tools, including protein identification tools, Blast, depending on the server !…. Not included yet in the UniProtKB complete proteome sets !

99. Depending upon the organism, the inclusion of alternative sequences to the basic set of protein sequences can make a tremedous difference. For instance, in Homo sapiens, alternative sequences currently represent close to 40% of the total number of annotated human sequences described in UniProtKB/Swiss-Prot. http://www.uniprot.org/faq/38

100. Blast P04150 against Swiss-Prot / homo sapiens @ UniProt Isoform sequences

101. Blast P04150 against Swiss-Prot / homo sapiens @ NCBI The isoform sequences are not present in the NCBI protein databases ! The.x number (P06401.4) correspond to the version number of the sequence…not to an alternatively spliced sequence !

102. How to track sequence changes ? The sequence version number applies to the canonical sequence only There is no easy way yet to track sequence updates of isoforms

103. UniProtKB 2- Biological data curation

104. UniProtKB/Swiss-Prot gathers data form multiple sources: - publications (literature/Pubmed) - prediction programs (Prosite, TMHMM, …) - contacts with experts - other databases - nomenclature committees An evidence attribution system allows to easily trace the source of each annotation Extract literature information and protein sequence analysis maximum usage of controlled vocabulary

105. Protein and gene names

106. …enable researchers to obtain a summary of what is known about a protein… General annotation (Comments) www.uniprot.org

107. Human protein manual annotation: some statistics (Aug 2010)

108. Sequence annotation (Features) …enable researchers to obtain a summary of what is known about a protein… www.uniprot.org

109. Human protein manual annotation: some statistics (PTM)

110. Non-experimental qualifiers UniProtKB/Swiss-Prot considers both experimental and predicted data and makes a clear distinction between both. Level. Type of evidenceQualifier 1st. Strong experimental evidenceRef.X 2nd. Light experimental evidenceProbable 3rd. Inferred by similarity with homologous protein (data of 1st or 2 nd level) By similarity 4th. Inferred by sequence predictionPotential

111. Find all the protein localized in the cytoplasm (experimentally proven) which are phosphorylated on a serine (experimentally proven)

112. UniProtKB Additional information can be found in the cross-references (to more than 140 databases)

113. 2D gel 2DBase-Ecoli ANU-2DPAGE Aarhus/Ghent-2DPAGE (no server) COMPLUYEAST-2DPAGE Cornea-2DPAGE DOSAC-COBS-2DPAGE ECO2DBASE (no server) OGP PHCI-2DPAGE PMMA-2DPAGE Rat-heart-2DPAGE REPRODUCTION-2DPAGE Siena-2DPAGE SWISS-2DPAGE UCD-2DPAGE World-2DPAGE Family and domain Gene3D HAMAP InterPro PANTHER Pfam PIRSF PRINTS ProDom PROSITE SMART SUPFAM TIGRFAMs Organism-specific AGD ArachnoServer CGD ConoServer CTD CYGD dictyBase EchoBASE EcoGene euHCVdb EuPathDB FlyBase GeneCards GeneDB_Spombe GeneFarm GenoList Gramene H-InvDB HGNC HPA LegioList Leproma MaizeGDB MGI MIM neXtProt Orphanet PharmGKB PseudoCAP RGD SGD TAIR TubercuList WormBase Xenbase ZFIN Protein family/group Allergome CAZy MEROPS PeroxiBase PptaseDB REBASE TCDB Genome annotation Ensembl EnsemblBacteria EnsemblFungi EnsemblMetazoa EnsemblPlants EnsemblProtists GeneID GenomeReviews KEGG NMPDR TIGR UCSC VectorBase Enzyme and pathway BioCyc BRENDA Pathway_Interaction_DB Reactome Other BindingDB DrugBank NextBio PMAP-CutDB Sequence EMBL IPI PIR RefSeq UniGene 3D structure DisProt HSSP PDB PDBsum ProteinModelPortal SMR PTM GlycoSuiteDB PhosphoSite PhosSite UniProtKB/Swiss-Prot: 129 explicit links and 14 implicit links! Proteomic PeptideAtlas PRIDE ProMEX PPI DIP IntAct MINT STRING Phylogenomic dbs eggNOG GeneTree HOGENOM HOVERGEN InParanoid OMA OrthoDB PhylomeDB ProtClustDB Polymorphism dbSNP Gene expression ArrayExpress Bgee CleanEx Genevestigator GermOnline Ontologies GO

114. Protein sequence origin http://www.uniprot.org/faq/35

115. Access to UniProtKB www.uniprot.org

116. The UniProt web site - www.uniprot.org Powerful search engine, google-like and easy-to-use, but also supports very directed field searches (similar to SRS) Scoring mechanism presenting relevant matches first Entry views, search result views and downloads are customizable The URL of a result page reflects the query; all pages and queries are bookmarkable, supporting programmatic access Tools: Blast, Align, IDmapping, Batch retrieval (Retrieve)

117. Search A very powerful text search tool with autocompletion and refinement options allowing to look for UniProt entries and documentation by biological information

118. Search A very powerful text search tool with autocompletion and refinement options allowing to look for UniProt entries and documentation by biological information

119. The search interface guides users with helpful suggestions and hints

121. Advanced Search A very powerful search tool To be used when you know in which entry section the information is stored Have first a look to annotation examples.

122. Find all human proteins with experimental evidence for their location in the nucleus

123. The information is stored in the ‘General annotation’ section, Subcellular location

125. Find all human proteins with experimental evidence for their location in the nucleus

126. Result pages: Highly customizable

127. Custom downloads…. Accession Genes Domains Protein Existence P02768 ALB (GIG20) (GIG42) (PRO0903) (PRO1708) (PRO2044) (PRO2619) (PRO2675) (UNQ696/PRO1341) Albumin domains (3) Evidence at protein level P02769 ALB Albumin domains (3) Evidence at protein level P02770 Alb Albumin domains (3) Evidence at protein level P07724 Alb (Alb-1) (Alb1) Albumin domains (3) Evidence at protein level P08759 alb-A Albumin domains (3) Evidence at transcript level P14872 alb-B Albumin domains (3) Evidence at transcript level P43652 AFM (ALB2) (ALBA) Albumin domains (3) Evidence at protein level P08835 ALB Albumin domains (3) Evidence at protein level P49822 ALB Albumin domains (3) Evidence at protein level P19121 ALB Albumin domains (3) Evidence at protein level Open with Excel etc.

128. The URL (results) can be bookmarked and manually modified.

129. Blast A tool associated with the standard options to search sequences in UniProt databases

130. Blast results: customize display

131. Blast: use of UniProt annotation amino-acids highlighting options and feature annotation highlighting option in the local alignment

132. Align A ClustalW multiple alignment tool with amino-acids highlighting options and feature annotation highlighting option

133. ClustalW multiple alignment of insulin sequences amino-acids highlighting options and feature annotation highlighting option in the local alignment

134. Retrieve A UniProt specific tool allowing to retrieve a list of entries in several standard identifiers formats. You can then query your ‘personal database’ with the UniProt search tool.

135. Your dataset: results of a Scan Prosite

136. ID Mapping Gives the possibility to get a mapping between different databases for a given protein

137. These identifiers are all pointing to TP53 (p53) ! P04637, NP_000537, ENSG00000141510, CCDS11118, UPI000002ED67, IPI00025087, etc.

139. Download

140. Downloading UniProt http://www.uniprot.org/downloads http://www.uniprot.org/downloads

141. Downloading UniProt http://www.uniprot.org/downloads http://www.uniprot.org/downloads

142. Canonical and isoform sequences

143. Complete proteome ‘gene’ centred or all known proteins ?

144. http://www.uniprot.org/faq/38

146. Remark: Some peptides are not associated with the keyword ‘Complete proteome’ because they do not match with the human genome

147. UniProt proteome sets, if downloaded in UniProt flat file or XML format, contain one sequence per UniProt record ! ‘gene’ centred all protein sequences in UniProtKB/Swiss-Prot… Are missing: other alternatively spliced protein sequences in UniProtKB/TrEMBL

148. IPI closure The complete MOUSE proteome will be composed of all MOUSE sequences in UniProtKB/Swiss-Prot plus those MOUSE sequences in UniProtKB/TrEMBL that have a cross- reference to an Ensembl protein. The complete HUMAN proteome will therefore be composed of all HUMAN sequences in UniProtKB/Swiss-Prot plus those HUMAN sequences in UniProtKB/TrEMBL that have a cross- reference to an Ensembl protein. News: 30th March 2011: next UniProtKB release.

149. UniProtKB Statistics

150. 520’000 + 14’000’000  14’000’000 Swiss-Prot & TrEMBL introduce a new arithmetical concept ! Redundancy in TrEMBL & Redundancy between TrEMBL and Swiss-Prot 12’000 species130’000 species Swiss-ProtTrEMBL

151. 12’000 species mainly model organisms

152. Not yet available

153. ~ 200 new entries / day new release every 4 weeks - Annotation is useful, good annotation is better, update is essential ! - Some entries have gone through more than 120 versions since their integration in UniProtKB/Swiss-Prot

154. UniProtKB entry history Always cite the primary accession number (AC) !

155. UniParc

156. - non-redundant protein sequence archive, containing both active and inactive sequences (including sequences which are not in UniProtKB i.e. immunoglobulins….) - the equivalent of ENA/GenBank/DDBJ at the protein level - species-merged: merge sequences between species when 100% identical over the whole length. - no annotation (only taxonomy) - can be searched only with database names, taxonomy, checksum (CRC64) and accession numbers (ACs) or UniProtKB, UniRef and UniParc IDs. - Beware: contains wrong prediction, pseudogenes etc…

157. Query UniParc

158. UniRef

159. ‘UniRef is useful for comprehensive BLAST similarity searches by providing sets of representative sequences’

160. «Collapsing BLAST results» Three collections of sequence clusters from UniProtKB and selected UniParc entries: One UniRef100 entry -> all identical sequences (identical sequences and sub-fragments are grouped in a single record) -> reduction of 12 % One UniRef90 entry -> sequences that have at least 90 % or more identity -> reduction of 40 % One UniRef50 entry -> sequences that are at least 50 % identical -> reduction of 65 % Based on sequence identity -> Independent of the species !

162. Independent of species and sequence length UniRef 90

163. UniMes

164. The UniProt Metagenomic and Environmental Sequences (UniMES) database is a repository specifically developed for metagenomic and environmental protein data (only GOS data for the moment). Download only (but included in UniParc -> Blast). - UniMES Fasta sequences - UniMES matches to InterPro methods ftp.uniprot.org/pub/databases/uniprot

166. UniMES: sequences in fasta format

168. Menu Introduction Nucleic acid sequencedatabases ENA/GenBank, DDBJ Protein sequence databases UniProt databases (UniProtKB) NCBI protein databases

169. NCBI protein databases (Entrez protein, NCBI nr) http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=Protein

170. Major ‘general’ protein sequence database ‘sources’ UniProtKB: Swiss-Prot + TrEMBL NCBI-nr: Swiss-Prot + TrEMBL + GenPept + PIR + PDB + PRF + RefSeq + TPA PIR PDBPRF UniProtKB/Swiss-Prot: manually annotated protein sequences (12’000 species) UniProtKB/TrEMBL: submitted CDS (ENA) + automated annotation; non redundant with Swiss-Prot (300’000 species) GenPept: submitted CDS (GenBank); redundant with Swiss-Prot (300’000 species ?) PIR: Protein Information Ressource; archive since 2003; integrated into UniProtKB PDB: Protein Databank: 3D data and associated sequences PRF: Protein Research Foundation journal scan of ‘published’ peptide sequences RefSeq: Reference Sequence for DNA, RNA, protein + gene prediction + some manual annotation (11’000 species) TPA: Third part annotation Integrated resources ‘cross-references’ Resources kept separated TPA not complete !!! (only entries created before 2007 ?)

171. Query at Entrez protein http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=Protein

172. Typical result of a query at « Entrez protein » RefSeq Swiss-Prot Genpept

175. A Swiss-Prot entry with the NCBI look

176. A TrEMBL entry with the NCBI look !!!!

177. GI number ‘GenInfo identifier’ number - In addition to an AC number specific from the original database, each protein sequence in the NCBInr database (included Swiss-Prot entry) has a GI number.

178. AC

179. GI number: ‘GenInfo identifier’ number - If the sequence changes in any way, a new GI number will be assigned: GI identifiers provide a mechanism for identifying the exact sequence that was used or retrieved in a given search. - A separate GI number is assigned to each protein translation (alternative products) - A Sequence Revision History tool is available to track the various GI numbers, version numbers, and update dates for sequences that appeared in a specific GenBank record: http://www.ncbi.nlm.nih.gov/entrez/sutils/girevhist.cgi

180. ID/AC mapping

182. http://www.ebi.ac.uk/Tools/picr/

183. GenPept Translation from annotated CDS in GenBank Contains all translated CDS annotated in GenBank/ENA/DDBJ sequences - equivalent to UniProtKB/TrEMBL, except that it is redundant with other databases (Swiss-Prot, RefSeq, PIR….)

184. GenPept: ‘translations from all annotated coding regions (CDS) in GenBank’

185. RefSeq Produced by NCBI and NLM http://www.ncbi.nlm.nih.gov/books/bookres.fcgi/handbook/ch18.pdf FAQ: http://www.ncbi.nlm.nih.gov/books/NBK50679/ http://www.ncbi.nlm.nih.gov/RefSeq/

186. The Reference Sequence (RefSeq) collection aims to provide a comprehensive, integrated, non-redundant set of sequences, including genomic DNA, transcript (RNA), and protein products, for major research organisms. Protein – mRNA – genomic sequence Also chromosomes, organelle genomes, plasmids, intermediate assembled genomic contigs, ncRNAs. - tighly linked to Entrez Gene (« interdependent curated resources »)

187. Example: NP_000790 Beware: NeXtProt accession number: NX_P00918

189. KW AC Taxonomy References

190. GenBank source and status Annotation and ontologies

191. Curated records

192. UniProtKB vs RefSeq

194. UniProtKB/Swiss-Prot merges all CDS available for a given gene and describes the sequence differences UniProtKB/Swiss-Prot P04150 (GCR_HUMAN):

195. RefSeq chooses one or several protein reference sequences for a given gene: they do not annotate the sequence differences. - If there is an alternative splicing event, there will be several distinct entries for a given gene Example: GCR_HUMAN GCR_HUMAN UniProtKB/Swiss-Prot 1 UniProtKB entry7 RefSeq entries cross-linked with

196. Protein feature annotation found in RefSeq - Conserved domains - Signal and mature petides - Propagation of a subset of features from Swiss-Prot.

197. PTM annotation Swiss-Prot vs RefSeq GCR_human

198. RefSeq statistics The numbers are not comparable: entries ‘sequence’ (RefSeq) vs entries ‘gene’ (UniProtKB/Swiss-Prot)

199. Summary UniProtKB vs NCBI protein

200. ENA/GenBank/DDBJRefSeq www.ncbi.nlm.nih.gov/RefSeq/ UniProt www.uniprot.org Protein and nucleotide dataGenomic, RNA and protein dataProtein data only Biological data added by the submitters (gene name, tissue…) Biological data annotated by curators, also found in the corresponding Entrez Gene entry Biological data annotated by curators (Swiss-Prot), within the entry Not curated Partially manually curated (‘reviewed’ entries) Manually curated in Swiss-Prot, not in TrEMBL Author submissionNCBI creates from existing data + gene prediction UniProt creates from existing data Only author can revise (except TPA) NCBI revises as new data emerge UniProt revises as new data emerge Multiple records for same loci common Single records for each molecule of major organisms Single records for each protein from one gene of major organisms (in Swiss-Prot, TrEMBL is redundant) Records can contradict each other Identification and annotation of discrepancy No limit to species included Limited to model organismsPriority (but not limited) to model organisms Data exchanged among INSDC members NCBI database; collaboration with UniProt UniProt database; collaboration with NCBI (RefSeq, CCDS)

201. PIR

202. PIR: the Protein Identification Resource PIR-PSD is no more updated, but exists as an archive

203. PDB

204. PDB (Protein Data Bank), 3D structure Contains the spatial coordinates of macromolecule atoms whose 3D structure has been obtained by X-ray or NMR studies Contains also the corresponding protein sequences *The PIR-NRL3D database makes the sequence information in PDB available for similarity searches and other tools Includes protein sequences which are mutated, chimearic etc… (created specifically to study the effect of a mutation on the 3D structure)

205. PDB: Protein Data Bank www.rcsb.org/pdb/ Managed by Research Collaboratory for Structural Bioinformatics (RCSB) (USA). Associated with specialized programs allow the visualization of the corresponding 3D structure (e.g., SwissPDB-viewer, Chime, Rasmol)). Currently there are ~68’000 structural data for about 15’000 different proteins, but far less protein family (highly redundant) !

206. PDB: example

207. Coordinates of each atom Sequence

208. Visualisation with Jmol

209. PRF Protein Research Foundation

210. http://www.genome.jp/dbget-bin/www_bfind?prf Looks for the peptide sequence described in publication (and which are not submitted in databases !!!)

212. Other protein databases

213. Ensembl http://www.ensembl.org/ Review http://nar.oxfordjournals.org/cgi/content/full/35/suppl_1/D610 Annotation pipeline http://www.genome.org/cgi/content/full/14/5/942 http://www.ensembl.org/ http://nar.oxfordjournals.org/cgi/content/full/35/suppl_1/D610 http://www.genome.org/cgi/content/full/14/5/942

214. - Ensembl: align the genomic sequences with all the sequences found in ENA, UniProtKB/Swiss-Prot, RefSeq and UniProtKB/TrEMBL (-> known genes) - Also do gene prediction (-> novel genes) Ensembl= UniProtKB + RefSeq + gene prediction - DNA, RNA and protein sequences available for several species. - Ensembl concentrates on vertebrate genomes, but other groups have adapted the system for use with plant, fungal and metazoa genomes.

217. Example of problem (derived from gene prediction pipeline) Ensembl completes the human ‘proteome’ by predicting/annotating missing genes according to orthologous sequences.. ID URAD_HUMAN Unreviewed; 171 AA. AC A6NGE7; DT 24-JUL-2007, integrated into UniProtKB/TrEMBL. DT 24-JUL-2007, sequence version 1. DT 02-OCT-2007, entry version 3. DE 2-oxo-4-hydroxy-4-carboxy-5-ureidoimidazoline decarboxylase homolog DE (OHCU decarboxylase homolog) (Parahox neighbour). GN Name=PRHOXNB; … DR EMBL; AL591024; -; NOT_ANNOTATED_CDS; Genomic_DNA. DR Ensembl; ENSG00000183463; Homo sapiens. DR HGNC; HGNC:17785; PRHOXNB. PE 4: Predicted; In primates the genes coding for the enzymes for the degradation of uric acid were inactivated and converted to pseudogenes.

218. IPI http://www.ebi.ac.uk/IPI/IPIhelp.html IPI: Closure ! http://www.ebi.ac.uk/IPI/IPIhelp.html

220. Automatic approach that builds clusters through combining knowledge already present in the primary data source (UniProtKB, RefSeq, Ensembl) and sequence similarity. IPI=UniProtKB + RefSeq + Ensembl (+ H-InvDB, TAIR +VEGA). !!! Complete proteome sets include all alternative splicing sequences…. Available for human, mouse, rat, Zebrafish, Arabidopsis, Chicken, and Cow

224. CCDS

225. http://www.ncbi.nlm.nih.gov/CCDS/

226. CCDS (human, mouse) Combining different approaches – ab initio, by similarity - and taking advantage of the expertise acquired by different institutes, including manual annotation… Consensus between 4 institutions…

228. Gene Ontology (GO)

229. Standards : Why is it so important ? ‘The ever-increasing number of sequencing projects necessitates a standardized system (…) to ensure that the flood of information produced can be effectively utilized.‘ (PMID 19577473 ) Standardization of biological data/information (data sharing and computational analysis). Aim: extract and compare annotation between different resources or species (semantic similarity).

230. Secreted or not secreted ? Pubmed19299134

231. The Gene Ontology is a controlled vocabulary, a set of standard terms—words and phrases—used for indexing and retrieving information. In addition to defining terms, GO also defines the relationships between the terms, making it a structured vocabulary. Contains ~30’000 terms. Gene Ontology (GO)

232. Gene Ontology (GO) terms  biological process broad biological phenomena e.g. mitosis, growth, digestion  molecular function molecular role e.g. catalytic activity, binding  cellular component Subcellular location e.g nucleus, ribosome, origin recognition complex

233. GO terms associated with human Erythropoietin

235. http://www.geneontology.org

236. Caveats Annotation is the process of assigning/mapping GO terms to gene products… Electronic vs Manual annotation…

237. Example with EPO

240. Histone H4 !!! Large scale derived data (‘proteome’)

241. GO terms: Essential link between biological knowledge and high throuput genomic and proteomic datasets… PMID: 15514041 ‘summary of the gene ontology classifications for all mapped ESTs…’

242. ~40 % of human proteins have no known function (experimental data)…but many more are associated with GO terms…(computer-assigned).

243. All documents (including practicals) are online http://education.expasy.org/cours/UniProt/