Textpresso-based automated extraction of concise descriptions

From WormBaseWiki
Jump to navigationJump to search

Querying for gene sets

Set of genes with a concise description

Query for all genes with a concise description from Postgres: Relevant postgres table names:

  • con_wbgene: Stores the WBGene ID and gene names
  • con_desctype: Type of description (relevant for us: Concise_description)
  • con_desctext: Text of the concise description

Query for all WBGenes that have a concise description (in con_desctext AND con_desctype):

SELECT DISTINCT(con_wbgene) FROM con_wbgene WHERE joinkey IN (SELECT joinkey FROM con_desctext WHERE con_desctext IS NOT NULL) AND joinkey IN (SELECT joinkey FROM con_desctype WHERE con_desctype IS NOT NULL) ORDER BY con_wbgene;

  • Number of genes with a concise description (as of 05.07.2014)=6,624

Set of genes with no concise description

Set of genes with no concise description and at least one published paper

Location of project-related files on Textpresso

http://textpresso-dev.caltech.edu/concise_descriptions/


Semantic categories in a Concise Description

1. Molecular identity
2. Orthology/Similarity
3. Mutant Phenotypes
4. Processes
5. Pathways
6. Genetic Interaction
7. Physical Interaction
8. Gene regulation data
9. Molecular Function
10. Tissue expression (may include life-stage)
11. Sub-cellular localization (may include life-stage)

Sources for mining data

1. Orthology, Homology and Paralog data

  • Ace tags: ?Gene Ortholog_other, Paralog
  • Contact: Michael Paulini

From Michael Paulini:

 
a.) orthology
it is in the ACeDB database on the genes as ortholog/paralog/ortholog_other, but we also dump it since a while each build (as   example, here: ftp://ftp.wormbase.org/pub/wormbase/releases/WS243/species/c_elegans/PRJNA13758/annotation 
/c_elegans.PRJNA13758.WS243.orthologs.txt.gz). 

b.) homology
1. protein homology)
the blastx data is in the GFF files, as well as here: ftp://ftp.wormbase.org/pub/wormbase/releases/WS243/acedb                 /Non_C_elegans_BLAST
for C.elegans the patch file is also loaded during the build, so you can find them as regular Homology_data on the respective  parent sequences in ACeDB.

the blastp data is as Homology_data on the proteins, as well as partially dumped into that one: ftp://ftp.wormbase.org/pub/wormbase/releases/WS243/species/c_elegans/PRJNA13758/c_elegans.PRJNA13758.WS243.best_blastp_hits.txt.gz

We also go protein clusters (mostly eggNOG based ones), which show homoloy and shared function and are connected to the member proteins in ACeDB through the Homology_group tag .

2. nucleotide homology) 
Mostlyy based on blat, but with the current release switched to star
You can find them in the respective GFF files and also similar to the blastx as homology data on the parent sequences in ACeDB

We also got RNASeq, which currently lives as RNASeq features in the GFF and ACeDB, but also as expression level data in the Gene/Transcript/CDS.

And last, but not least, we got pairwise whole genome alignments for selected species, which currently we only show on EnsEMBL Genomes, but you can use the generic Compara API to pull the alignments from there.

As orthology + homology covers such a huge swath of very different data in WormBase, there is no unifying format, except ACeDB and to a certain extent GFF.

2. Process

  • Caltech data
  • GO data
  • Source 1: Download gene_association.wb.gz for C. elegans from GO consortium: http://www.geneontology.org/GO.current.annotations.shtml?all
    • All rows with column 15 (assigned by) with 'WB' are WormBase annotations, those with 'UniProtKB' or 'InterPro' are from those databases
    • Need data from these rows:
      • where column 9: has value 'P' (Process),
      • column 2 (DB_Object ID): the associated genes, UniProt ID or WBGene ID,
      • column 3 (DB_Object symbol), eg, wht-7, GO terms are from column 5.
      • column 5: GOID, eg, GO:0000346
      • column 6: DB:Reference (Reference), eg.PMID:12062106, GO_REF:0000002
      • column 7: Evidence code, eg, IMP
      • column 8: With, eg. 'WB:WBRNAi00000785|WBPhenotype:0000050'
  • GO data in postgres
    • Paper -- gop_paper
    • WBGene -- gop_wbgene
    • GO -- gop_goontology
    • GO Term -- gop_goid
  • Contact Person: Kimberly, Ranjana
  • Source 2: Topic data
    • OA field: Gene, PG table name:pro_wbgene
    • OA field:WBPaper, PG table name:pro_paper
  • Contact Person: Karen


3. Molecular identity

  • Caltech and non-caltech data?
  • Source 1: GO data
    • Download the gene_association file for C.elegans from the WormBase FTP site:
    • ftp://ftp.sanger.ac.uk/pub/wormbase/releases/WS243/ONTOLOGY/gene_association.WS243.wb.c_elegans
    • All rows with column 15 (assigned by) with 'WB' are WormBase annotations, those with 'UniProtKB' or 'InterPro' are from those databases
    • Need data from these rows:
      • where column 9 has value 'F' (Molecular Function)
      • column 2, associated genes, has UniProt ID or WormBase GeneID, need to translate UniProtID to WormBase ID.
      • column 3: DB_Object symbol, eg, wht-7,
      • column 5: GOID, eg, GO:0000346
      • column 6: DB:Reference (Reference), eg.PMID:12062106, GO_REF:0000002
      • column 7: Evidence code, eg, IMP
      • column 15: Assigned By, eg., WB (which database created the annotation)

4. Pathway (No database source for now?)

5. Mutant Phenotype

  • Caltech data
  • Source 1: Phenotype OA, PG table name:(Phenotypes are added to variation and not genes)
  • Source 2: Acedb tag: Under ?Gene, Reference_allele ?Variation and Allele ?Variation and Under ?Variation Phenotype
  • Source 3: phenotype_association file:ftp://ftp.wormbase.org/pub/wormbase/releases/WS243/ONTOLOGY/
    • File name:phenotype_association.WS243.wb
    • Following rows:
      • Rows with 'NOT' should be ignored for this search.
      • column 2, associated genes, WormBase GeneID
      • column 3: DB_Object symbol, eg, aap-1,
      • column 5: Phenotype ID, eg, WBPhenotype:0000674
      • column 6: DB:Reference (Reference), eg.WB_REF:WBPaper00032243, or WB:WBVar00249743
  • Contact Person: Karen

6. Genetic Interaction and 7. Physical Interaction

  • Caltech data
  • Source 1: gene_association.wb
    • ftp://ftp.wormbase.org/pub/wormbase/releases/WS243/ONTOLOGY/
    • File name:gene_association.WS243.wb.c_elegans
    • Rows with a 'IGI' in column 7 indicate a genetic interaction between the WBgenes in column 2/3 and column 8
    • Rows with a 'IPI" in column 7 indicate a physical interaction between the WBgenes in column 2/3 and column 8
  • Source 2: Interaction OA and tables
    • "Field Name" = Postgres Table:
    • "Paper" = int_paper
    • "Interaction Type" = int_type
    • "Bait overlapping gene" = int_genebait
    • "Target overlapping gene" = int_genetarget
    • "Non-directional Gene(s)" = int_genenondir
    • "Effector Gene(s)" = int_geneone
    • "Affected Gene(s)" = int_genetwo

Example statements:

If int_type = "Physical"

<int_genebait> interacts physically with <int_genetarget> (and vice versa)


If int_type = "Genetic - Synthetic ( Synthetic )"

<int_genenondir> interacts with <other int_genenondir(s)> in a synthetic genetic interaction


If int_type = "Genetic - Suppression ( Suppression )"

<int_geneone> genetically suppresses <int_genetwo>

8. Gene regulation

  • Caltech data
  • Source: Gene regulation data in genereg OA
    • Positive_regulate Anatomy_term "<grg_pos_anatomy>"
    • Positive_regulate Life_stage "<grg_pos_lifestage>"
    • Positive_regulate Subcellular_localization "<grg_pos_scl>"
    • Positive_regulate Subcellular_localization_text "<grg_pos_scltext>"
    • Negative_regulate Anatomy_term "<grg_neg_anatomy>"
    • Negative_regulate Life_stage "<grg_neg_lifestage>"
    • Negative_regulate Subcellular_localization "<grg_neg_scl>"
    • Negative_regulate Subcellular_localization_text "<grg_neg_scltext>"
    • Does_not_regulate Anatomy_term "<grg_not_anatomy>"
    • Does_not_regulate Life_stage "<grg_not_lifestage>"
    • Does_not_regulate Subcellular_localization "<grg_subcellloc>"
    • Does_not_regulate Subcellular_localization_text "<grg_not_scltext>"
    • Trans_regulated_gene "<grg_transregulated>"
    • Trans_regulator_gene "<grg_transregulator>"
    • No Subdata Result "<grg_result>"

9. Molecular Function/Identity

  • Caltech data: GO Molecular Function
  • Source 1: Download gene_association.wb.gz for C. elegans from GO consortium: http://www.geneontology.org/GO.current.annotations.shtml?all
    • All rows with column 15 (assigned by) with 'WB' are WormBase annotations, those with 'UniProtKB' or 'InterPro' are from those databases
    • Need data from these rows:
      • where column 9: has value 'F' (Molecular Function),
      • column 2 (DB_Object ID): the associated genes, UniProt ID or WBGene ID,
      • column 3 (DB_Object symbol), eg, wht-7, GO terms are from column 5.
      • column 5: GOID, eg, GO:0000346
      • column 6: DB:Reference (Reference), eg.PMID:12062106, GO_REF:0000002
      • column 7: Evidence code, eg, IDA
  • Contact Person: Kimberly, Ranjana

10. Tissue expression and life stage

  • Caltech data
  • Source 1: Expression data
  • OA (exprpat), PG table names:
    • exp_anatomy for anatomy terms
    • exp_goid for subcell localization
    • exp_lifestage for life stage
    • exp_paper for paper
    • exp_gene for gene
  • Contact Person: Daniela

11. Sub-cellular localization

  • Caltech data
  • GO data
  • Source 1: gene_association file for C. elegans at:http://www.geneontology.org/GO.current.annotations.shtml?all
    • Need data from these rows:
      • where column 9 has value 'C' (Cellular Component)
      • column 2, associated genes, has UniProt ID or WormBase GeneID, need to translate UniProtID to WormBase ID.
      • column 3: DB_Object symbol, eg, wht-7,
      • column 5: GOID, eg, GO:0000346
      • column 7: Evidence code, eg, IDA

Template for a Concise Description

For the test phase, order of sentences:

  • Orthology
  • Process
  • Function/identity
  • Component


Orthology/Similarity
<Gene> encodes an ortholog of <human protein>;
Phenotypes
<Gene> mutants exhibit the following phenotypes: <phenotypes>.
Process/Pathway
<Gene> is (required, functions, regulates, is involved in, is part of) <process>;
Genetic interaction with respect to Process or Pathway
<Gene> interacts genetically with <gene1, gene2>;
Physical interaction
<Protein> physically interacts with (protein, DNA, RNA) .....;
Molecular Function
<Protein> has <molecular function>..... activity in (in vitro, in vivo);
Tissue Expression and sub-cellular localization
<Gene/Protein> is expressed in <tissue> and localizes to <GO cellular component>;

Note: Not all descriptions may follow the exact order or choice of words, see Rules below.

Rules for automated sentence construction

Homology

<Gene> encodes an ortholog of human <human protein name>;

Process

  • Rule 1: Ignore all IEA and ISS process terms
  • Rule 2: Exclusions:
    • Ignore the term 'reproduction'[IMP]
    • Ignore the term 'embryo development ending in birth or egg hatching[IMP]'
  • Rule 3: If a GO term has the words 'involved in' anywhere, beginning, middle or at the end, use the words 'functions in'
    • Data: WBGene00006495,cpna-1,striated muscle contraction involved in embryonic body morphogenesis[IMP],WB_REF:WBPaper00041875|PMID:23283987
    • Sentence: cpna-1 functions in striated muscle contraction involved in embryonic body morphogenesis;
  • Rule 4: For the GO term 'synaptic transmission, <word>' switch the order of words to make it '<word> synaptic transmission'.
  • Example for Rule 4:
    • WBGene00011488,nra-2, synaptic transmission, cholinergic[IMP],WB_REF:WBPaper00034730|PMID:19609303,,WB,locomotion[IMP],protein processing[IEA],PMID:12520011|PMID:12654719,INTERPRO:IPR008710,regulation of signal transduction[IEA],INTERPRO:IPR016574
    • nra-2 is involved in cholinergic synaptic transmission and locomotion.
  • Rule 5: For all other Process terms the sentence will be:
    • <Gene> is involved in <process term>;
    • Examples:
    • WBGene00014123,elpc-3,tRNA wobble uridine modification[IMP],WB_REF:WBPaper00034713|PMID:19593383,,WB,translation[IMP],spermatogenesis[IGI],olfactory learning[IMP],vulval development[IGI],embryonic morphogenesis[IGI],oocyte development[IGI]
    • Sentence: elpc-3 is involved in tRNA wobble uridine modification, translation, spermatogenesis, olfactory learning, vulval development and embryonic morphogenesis.
  • Rule 6: For the term 'molting cycle, collagen and cuticulin-based cuticle' GO term, drop the comma and the words 'collagen and cuticulin-based cuticle'
    • Example:
    • WBGene00016643,vps-45,molting cycle, collagen and cuticulin-based cuticle[IMP],WB_REF:WBPaper00029049|PMID:17235359,,WB,vesicle docking involved in exocytosis[IEA],PMID:12520011|PMID:12654719,INTERPRO:IPR001619,vesicle-mediated transport[IEA]
    • Sentence: vps-45 is involved in the molting cycle;

Molecular identity/function

  • Rule 1: Exclusion list:
    • Ignore the term 'protein binding'
    • Ignore the term 'binding'
  • Rule 2: If the evidence code is 'IEA' and 'activity' term is present, add the words 'predicted to have <activity term>' and add the words ', based on protein domain information' to the sentence:
    • Sentence: <Gene> is predicted to have <activity term>, based on protein domain information.
  • Examples:
    • WBGene00000108,alh-2,oxidoreductase activity[IEA],PMID:12520011|PMID:12654719,INTERPRO:IPR015590,WB,oxidoreductase activity, acting on the aldehyde or oxo group of donors, NAD or NADP as acceptor[IEA],INTERPRO:IPR016163
    • alh-2 is predicted to have oxidoreductase activity, acting on the aldehyde or oxo group of donors, NAD or NADP as acceptor, based on protein domain information.
  • Rule 3: If Evidence code is 'ISS' add the words 'based on sequence information' to the sentence
    • WBGene00001951,hlh-4,sequence-specific DNA binding transcription factor activity[ISS],WB_REF:WBPaper00034761|PMID:19632181,,WB,protein dimerization activity[IEA],PMID:12520011|PMID:12654719,INTERPRO:IPR011598,DNA binding[IEA],INTERPRO:IPR015660
    • Sentence: hlh-4 is predicted to have sequence-specific DNA binding transcription factor activity based on sequence information, protein dimerization activity and DNA binding activity, based on protein domain information.
  • Rule 4: If a binding term is present add the word 'activity' to it.
  • Rule 5: If evidence code is 'IDA' or 'IMP' add the words 'exhibits' and to the sentence.
    • IDA example:
    • WBGene00001952,hlh-6,RNA polymerase II distal enhancer sequence-specific DNA binding transcription factor activity[IDA],WB_REF:WBPaper00032277|PMID:18927627,,WB,protein dimerization activity[IEA],PMID:12520011|PMID:12654719,INTERPRO:IPR011598,DNA binding[IEA],INTERPRO:IPR015660
    • Sentence: hlh-6 exhibits RNA polymerase II distal enhancer sequence-specific DNA binding transcription factor activity and is predicted to have protein dimerization activity and DNA binding activity.
    • IMP example:
    • WBGene00009583,aagr-3,alpha-glucosidase activity[IMP],WB_REF:WBPaper00036069|PMID:20349118,,WB,hydrolase activity, hydrolyzing O-glycosyl compounds[IEA],PMID:12520011|PMID:12654719,INTERPRO:IPR000322,catalytic activity[IEA],INTERPRO:IPR011013,carbohydrate binding[IEA]
    • Sentence: aagr-3 exhibits alpha-glucosidase activity and is predicted to have hydrolase activity, hydrolyzing O-glycosyl compounds, catalytic activity, and carbohydrate binding activity, based on protein domain information.

Rule 6: For terms beginning with 'structural constituent of <word>', use the words 'is a' in the sentence.

  • Example:
    • WBGene00010783,mrpl-36,structural constituent of ribosome[IEA],PMID:12520011|PMID:12654719,INTERPRO:IPR000473,WB
    • Sentence: mrpl-36 is a structural constituent of ribosome, based on protein domain information.
    • WBGene00010783,mrpl-36,structural constituent of ribosome[ISS]
    • Sentence: mrpl-36 is a structural constituent of ribosome, based on sequence information.
    • WBGene00010783,mrpl-36,structural constituent of ribosome[IMP] or [IDA]
    • Sentence: mrpl-36 is a structural constituent of ribosome, based on experimental evidence.

Sub-cellular localization/Component

  • Rule 1: Ignore all IEA and ISS GO terms, use only non-IEA GO terms
    • Sentence: <Gene> is localized to <component term>;
  • Rule 2: For 'integral component of ....' terms add the words 'is an';
    • Eg. for Rule 2: WBGene00006319,sup-10,integral component of plasma membrane[IDA],WB_REF:WBPaper00006135|PMID:14534247,,WB,striated muscle dense body[IDA]
    • sup-10 is an integral component of plasma membrane and is localized to striated muscle dense body;
  • Examples
    • WBGene00023405,sor-1,nucleoplasm[IDA],WB_REF:WBPaper00027128|PMID:16501168,,WB,nuclear speck[IDA]
    • Sentence: sor-1 is localized to the nucleoplasm and nuclear speck;
    • WBGene00004681,rsd-2,nucleolus[IDA],WB_REF:WBPaper00044261|PMID:18430922,,WB,endoplasmic reticulum[IDA],cytosol[IDA]
    • Sentence: rsd-2 is localized to the nucleolus, endoplasmic reticulum and cytosol;
    • WBGene00021827,dnc-6,dynactin complex[IDA],WB_REF:WBPaper00037699|PMID:20964796,,WB
    • Sentence: dnc-6 is localized to the dynactin complex;

Publications related to Text-mining methods

  • Automatically generating gene summaries from biomedical literature.

Ling X, Jiang J, He X, Mei Q, Zhai C, Schatz B.

Pac Symp Biocomput. 2006:40-51.

PMID:17094226

  • Generating gene summaries from biomedical literature: A study of semi-structured summarization

Xu Ling *, Jing Jiang, Xin He, Qiaozhu Mei, Chengxiang Zhai, Bruce Schatz

Information Processing and Management 43 (2007) 1777–1791


Back To Concise Descriptions