load-japonicus-chado.yaml

# PomBase Chado load config file  -*- conf -*-

# used for GAF export, for feature_cvterm and feature_relationship
# creation
#  - it's added as an "assigned_by" feature_cvtermprop
#  - source_database prop feature_relationship
database_name: JaponicusDB

cvs:
  cvterm_property_type:
    - nested_extension
    - extension_relations_status
    - malacards_disease_name
    - malacards_displayed_disease_name
    - malacards_disease_slug
  PomBase publication types:
    - paper
    - null
  PomBase chadoprop types:
    - db_creation_datetime
    - date_version
    - db_date_version
    - UniProt-GOA version
    - InterPro version
    - BioGRID version
  synonym_type:
    - reserved_name
    - obsolete_name
  PomBase feature property types:
    - name: comment
      definition: >
        A comment or note about the feature, similar to a /note qualifier in
        EMBL format files.
    - symbol
    - description
    - allele_type
    - uniprot_identifier
    - name: biogrid_interactor_id
      definition: the ID of this gene in the BioGRID database
    - canto_session
    - name: ena_id
      definition: The European Nucleotide Archive ID of the feature.
    - genotype_background
    - genotype_display_name
    - genotype_comment
    - molecular_weight
    - average_residue_weight
    - charge_at_ph7
    - isoelectric_point
    - codon_adaptation_index
    - name: sgd_identifier
      definition: "An ID like SGD:S000005390"
    - name: partial_sequence
      definition: This qualifier is present if the source feature has the EMBL /partial qualifer.
    - name: annotation_date
      definition: The date that this feature was added to the database.
    - name: kegg_pathway
    - name: rnacentral_identifier
  PomBase allele types:
    - name: amino_acid_deletion_and_mutation
    - name: amino_acid_insertion
    - name: amino_acid_insertion_and_deletion
    - name: amino_acid_insertion_and_mutation
    - name: amino_acid_mutation
    - name: deletion
    - name: disruption
    - name: nonsense_mutation
    - name: nucleotide_insertion
    - name: nucleotide_mutation
    - name: partial_nucleotide_deletion
    - name: wild_type
    - name: partial_amino_acid_deletion
      description: Deletion of part of a gene.
    - name: unrecorded
      description: An allele whose type is not apparent from the paper.
    - name: unknown
      description: An allele whose type has not been established.
    - name: other
    - name: unspecified
  PomBase annotation extension terms: []
  PomBase gene products: []
  PomBase family or domain: []
  PomBase gene characterisation status:
    - name: published
      definition: >
        A gene product characterised, or partially characterised, in a small
        scale experiment to provide published information about the biological
        process. The biological process annotation must be equivalent to, or
        more specific than, terms in the "GO BP slim"
        https://www.pombase.org/browse-curation/fission-yeast-bp-go-slim-terms
    - name: biological role inferred
      definition: >
        A gene product whose biological process is inferred from homology to an
        experimentally characterised gene product. The biological process
        annotation must be equivalent to, or more specific than, terms in the "GO
        BP slim"  https://www.pombase.org/browse-curation/fission-yeast-bp-go-slim-terms
    - name: conserved unknown
      definition: >
        A gene product which is conserved more broadly than the
        Schizosacchaomyces clade, but cannot be assigned a biological process
        term equivalent to, or more specific than, the terms in the "GO BP slim"
        https://www.pombase.org/browse-curation/fission-yeast-bp-go-slim-terms based on
        experimental data from any organism
    - name: Schizosaccharomyces pombe specific protein, uncharacterized
      definition: >
        A gene product which is not conserved outside the Schizosacchaomyces clade
        and where nothing is published about, or can be inferred from homology
        about, the biological role
    - name: Schizosaccharomyces specific protein, uncharacterized
      definition: >
        A gene product which is not conserved outside the Schizosacchaomyces clade
        and where nothing is published about, or can be inferred from homology
        about, the biological role
    - name: dubious
      definition: Unlikely to be coding
    - name: transposon
      definition: >
        A transposon or insertion sequence. An element that can insert
        in a variety of DNA sequences.
  PomBase interaction types:
    - name: interacts_genetically
    - name: interacts_physically
  feature_relationshipprop_type:
    - name: evidence
    - name: source database
    - name: source file
    - name: homology type
    - name: ortholog qualifier
    - name: first_approved_timestamp
    - name: approved_timestamp
    - name: approver_email
    - name: curator_name
    - name: curator_email
    - name: canto_session
    - name: interaction_note
    - name: is_inferred
      definition: >
        Set to "yes" for annotation inferred during loading, "no" otherwise.
  pubprop_type:
    - name: canto_session
    - name: canto_first_sent_to_curator_date
    - name: canto_first_approved_date
    - name: canto_annotation_status
    - name: canto_approved_date
    - name: canto_approver_name
    - name: canto_approver_email
    - name: canto_initial_curator_name
    - name: canto_initial_curator_email
    - name: canto_curator_name
    - name: canto_curator_email
    - name: previous_curator_name
    - name: canto_curator_role
    - name: canto_triage_status
    - name: canto_added_date
    - name: canto_session_accepted_date
    - name: canto_session_submitted_date
    - name: canto_message_for_curators
    - name: canto_experiment_type
    - name: pubmed_publication_date
    - name: pubmed_electronic_publication_date
    - name: pubmed_entrez_date
    - name: pubmed_authors
    - name: pubmed_citation
    - name: pubmed_abstract
    - name: pubmed_doi
    - name: annotation_curator
  cat_act: []
  genome_org: []
  m_f_g: []
  DNA_binding_specificity: []
  name_description: []
  species_dist: []
  subunit_composition: []
  gene_ex:
    - RNA level
    - protein level
    - transcription
    - RNA degradation
    - translation
    - protein degradation
  complementation: []
  warning: []
  ex_tools: []
  external_link: []
  misc: []
  disease_associated: []
  EC numbers: []

# define the allowed extension relation for each annotation type
extension_restrictions:
  fission_yeast_phenotype:
    allowed:
      - is_bearer_of
      - has_penetrance
      - assayed_using
      - has_severity
      - occurs_in
      - occurs_at
      - assayed_enzyme
      - assayed_substrate
      - assayed_region
      - assayed_protein
      - assayed_gene
  biological_process:
    allowed:
      - happens_during
      - has_input
      - occurs_at
      - occurs_in
      - has_regulation_target
      - part_of
      - not_happens_during
      - regulates_activity_of
      - directly_positively_regulates
      - directly_negatively_regulates
  molecular_function:
    allowed:
      - happens_during
      - has_input
      - has_direct_input
      - has_substrate
      - occurs_at
      - occurs_in
      - has_regulation_target
      - in_presence_of
      - activated_by
      - inhibited_by
      - part_of
      - not_happens_during
      - directly_positively_regulates
      - directly_negatively_regulates
      - regulates_activity_of
  cellular_component:
    allowed:
      - exists_during
      - part_of
      - coincident_with
      - not_exists_during
  'PSI-MOD':
    allowed:
      - absent_during
      - added_by
      - added_during
      - affected_by
      - coincident_with
      - decreased_during
      - has_direct_input
      - in_presence_of
      - increased_during
      - level_fluctuates_during
      - occupancy
      - present_during
      - removed_by
      - removed_during
      - required_for
  gene_ex:
    allowed:
      - during
      - in_absence_of
      - in_presence_of
      - occurs_in
  PomGeneExRNA:
    allowed:
      - during
      - in_absence_of
      - in_presence_of
      - occurs_in
  PomGeneExProt:
    allowed:
      - during
      - in_absence_of
      - in_presence_of
      - occurs_in
  PomGeneExRD:
    allowed:
      - during
      - in_absence_of
      - in_presence_of
  name_description:
    allowed: []
  subunit_composition:
    allowed:
      - located_in


# term IDs where duplicate extensions are allowed
allowed_duplicate_extensions:
  - 'FYPO:0000702'
  - 'FYPO:0000703'
  - 'FYPO:0000704'
  - 'FYPO:0000705'
  - 'FYPO:0001275'
  - 'FYPO:0001571'
  - 'FYPO:0001645'
  - 'FYPO:0002365'
  - 'FYPO:0003206'
  - 'FYPO:0003207'
  - 'FYPO:0003215'
  - 'FYPO:0004335'

# names of the CVs that extension relation terms can come from
extension_relation_cv_names:
  - relationship
  - go/extensions/gorel
  - fypo_extension_relations
  - PSI-MOD_extension_relations
  - external
  - fission_yeast_phenotype

extension_relation_transform:
  cellular_component:
    happens_during: exists_during
    occurs_during: exists_during
    during: exists_during

cv_definitions:
  biological_process: Gene Ontology biological process terms
  cat_act: Catalytic activity attributes (Kd, rate, etc.)
  complementation: gene product complement/complemented by another gene product
  disease_associated: Orthology to a human disease gene
  DNA_binding_specificity: Target sequence consensus of a DNA binding protein
  ex_tools: Uses of gene products as experimental tools (e.g. constructs, selectable markers, antibodies, etc.)
  gene_ex: GP annotations related to gene expression
  genome_org: Genome organization attributes (e.g. tandem repeat)
  m_f_g: Miscellaneous functional groups, not captured by GO (e.g. chaperone)
  misc: Annotation which does not fit any other ontology
  name_description: Derivation of the "three letter" gene name
  PomBase annotation extension terms: Post-composed terms used for annotation cross products
  PomBase family or domain: Manual annotation of Pfam false negatives and uncaptured subfamilies
  PomBase gene products: Concise gene product names
  species_dist: Broad classifiers of taxon distribution (domain/kingdom level)
  subunit_composition: Stoichiometry of homogeneous complexes
  warning: Alerts, gross changes and anomalies
  PSI-MOD: Protein modification terms

dbs:
  - RFAM
  - GOA
  - UniProtKB/Swiss-Prot
  - UniProtKB/TrEMBL
  - "EnsemblGenomes-Gn"
  - "EnsemblGenomes-Tr"
  - "UniProtKB/Swiss-Prot"
  - GOA
  - UniProt
  - SPD
  - PMID
  - EMBL
  - GOC
  - GO_REF
  - KEGG
  - Pfam
  - Rfam
  - GI
  - InterPro
  - UniProtKB
  - UniProtKB/Swiss-Prot
  - TreeFam

# where to store the identifiers from each /db_xref we see
db_dest_tables:
  UniProtKB: dbxref
  "UniProtKB/TrEMBL": dbxref
  "EnsemblGenomes-Gn": dbxref
  "EnsemblGenomes-Tr": dbxref
  "UniProtKB/Swiss-Prot": dbxref
  PDB: dbxref
  GOA: dbxref
  Rfam: dbxref
  InterPro: dbxref
  PMID: pub
  SPD: dbxref
  KEGG: dbxref
  EMBL: pub

taxonid: 4897

organism_prefixes:
  Schizosaccharomyces_pombe: PomBase

allowed_unknown_term_names_files:
  - /var/pomcur/sources/pombe-embl/chado_load_mappings/cumulative_unknown_term_name_ok.txt
allowed_term_mismatches_files:
  - /var/pomcur/sources/pombe-embl/chado_load_mappings/cumulative_ID_ok_mismatches.txt
  - /var/pomcur/sources/pombe-embl/chado_load_mappings/cumulative_name_ok_mismatches.txt

# This allows species to merged into a parent taxa when loading.  For
# example when loading annotation from GOA into PomBase we can merge
# taxon 284812 (Schizosaccharomyces pombe 972h-) into 4896
# (Schizosaccharomyces pombe) rather than having two organisms for pombe
# in Chado
organism_taxon_map:
  402676: 4897
  1248645: 4897
  284812: 4896
  559292: 4932

# a Perl regular expression that matches the systematic ID / uniquenames
# in this database
systematic_id_re: 'SJAG_\d+'

# the default is to use the cv name as the db name
db_names_for_cvs:
  biological_process: GO
  molecular_function: GO
  cellular_component: GO
  sequence: SO

geneontology_qualifier_flags:
  NOT: 1
  contributes_to: 1
  colocalizes_with: 1

evidence_types:
  IMP:
    name: Inferred from Mutant Phenotype
    throughput_type: low throughput
  IDA:
    name: Inferred from Direct Assay
    throughput_type: low throughput
  IGI:
    name: Inferred from Genetic Interaction
    throughput_type: low throughput
    with_gene: 1
  IPI:
    name: Inferred from Physical Interaction
    throughput_type: low throughput
    with_gene: 1
  EXP:
    name: Inferred from Experiment
    throughput_type: low throughput
  IEP:
    name: Inferred from Expression Pattern
    throughput_type: non-experimental
  ISS:
    name: Inferred from Sequence or Structural Similarity
    throughput_type: non-experimental
  ISO:
    name: Inferred from Sequence Orthology
    throughput_type: non-experimental
  ISA:
    name: Inferred from Sequence Alignment
    throughput_type: non-experimental
  ISM:
    name: Inferred from Sequence Model
    throughput_type: non-experimental
  IGC:
    name: Inferred from Genomic Context
    throughput_type: non-experimental
  IBA:
    name: Inferred from Biological aspect of Ancestor
    throughput_type: non-experimental
  IBD:
    name: Inferred from Biological aspect of Descendant
    throughput_type: non-experimental
  IKR:
    name: Inferred from Key Residues
    throughput_type: non-experimental
  IRD:
    name: Inferred from Rapid Divergence
  RCA:
    name: inferred from Reviewed Computational Analysis
  NAS:
    name: Non-traceable Author Statement
    throughput_type: non-experimental
  IC:
    name: Inferred by Curator
    throughput_type: non-experimental
  ND:
    name: No biological Data available
  IEA:
    name: Inferred from Electronic Annotation
    throughput_type: non-experimental
  NR:
    name: Not Recorded
  TAS:
    name: Traceable Author Statement
    throughput_type: non-experimental
  HTP:
    name: Inferred from high throughput experiment
    throughput_type: high throughput
  HDA:
    name: Inferred from high throughput direct assay
    throughput_type: high throughput
  HEP:
    name: Inferred from high throughput expression pattern
    throughput_type: high throughput
  HGI:
    name: Inferred from high throughput genetic interaction
    throughput_type: high throughput
  HMP:
    name: Inferred from high throughput mutant phenotype
    throughput_type: high throughput
  UNK:
    name: Unknown
  Microscopy: ~
  'Other': ~
  'Other direct assay': ~
  'Affinity Capture-Luminescence': ~
  'Affinity Capture-MS': ~
  'Affinity Capture-RNA': ~
  'Affinity Capture-Western': ~
  'Biochemical Activity': ~
  'Chromatography evidence': ~
  'Co-crystal Structure':
    name: Co-crystal Structure
    symmetrical: yes
  'Co-fractionation':
    name: Co-fractionation
    symmetrical: yes
  'Co-localization':
    name: Co-localization
    symmetrical: yes
  'Co-purification':
    name: Co-purification
    symmetrical: yes
  'Dosage Growth Defect': ~
  'Dosage Lethality': ~
  'Dosage Rescue': ~
  'FRET':
    name: FRET
  'Far Western': ~
  'Microarray RNA expression level data': ~
  'Negative Genetic':
    name: Negative Genetic
    symmetrical: yes
  'PCA':
    name: PCA
    symmetrical: yes
  'Phenotypic Enhancement': ~
  'Phenotypic Suppression': ~
  'Positive Genetic':
    name: Positive Genetic
    symmetrical: yes
  'Protein-RNA': ~
  'Protein-peptide': ~
  'Reconstituted Complex':
    name: Reconstituted Complex
    symmetrical: yes
  'Synthetic Growth Defect':
    name: Synthetic Growth Defect
    symmetrical: yes
  'Synthetic Haploinsufficiency':
    name: Synthetic Haploinsufficiency
    symmetrical: yes
  'Synthetic Lethality':
    name: Synthetic Lethality
    symmetrical: yes
  'Synthetic Rescue': ~
  'Two-hybrid': ~
  'Enzyme assay data': ~
  'Flow cytometry data': ~
  'Chromatin immunoprecipitation experiment': ~
  'Immunolocalization experiment data': ~
  'Epitope-tagged protein immunolocalization experiment data': ~
  'Protein localization experiment': ~
  'Co-immunoprecipitation experiment': ~
  'Electrophoretic mobility shift assay data': ~
  'Particle size and count assay': ~
  'Western blot assay': ~
  'Substance quantification': ~
  'Substance quantification evidence': ~
  'Plasmid maintenance assay evidence': ~
  'FPR':
    name: fitness profiling
    definition: "ECO term name is 'competitive growth assay evidence' (ECO:0000336)"
  'ECO:0000000':
    name: evidence
    definition: A type of information that is used to support an assertion.
  'ECO:0000007':
    name: immunofluorescence evidence
  'ECO:0000049':
    name: reporter gene assay evidence
  'ECO:0000106':
    name: Northern assay evidence
  'ECO:0000112':
    name: Western blot evidence
  'ECO:0000124':
    name: fusion protein localization evidence
  'ECO:0000058':
    name: expression microarray evidence
  'ECO:0000336':
    name: competitive growth assay evidence
  'ECO:0000108':
    name: reverse transcription polymerase chain reaction transcription evidence
  'ECO:0000110':
    name: RNA protection assay evidence
  'ECO:0000048':
    name: transcript expression level evidence
  'ECO:0000231':
    name: quantitative PCR
    definition: "ECO term name is 'quantitative polymerase chain reaction evidence'"
  'ECO:0000006':
    name: experimental evidence
  'ECO:0000337':
    name: gel electrophoresis evidence
  'ECO:0000221':
    name: high throughput nucleotide sequencing assay evidence
  'ECO:0000095':
    name: cell growth assay evidence
  'ECO:0000098':
    name: in situ hybridization assay evidence
  'ECO:0001096':
    name: mass spectrometry evidence
  'ECO:0000059':
    name: experimental phenotypic evidence
  'ECO:0000333':
    name: sodium dodecyl sulfate polyacrylamide gel electrophoresis evidence
  'ECO:0000279':
    name: western blot evidence used in manual assertion
  'ECO:0005653':
    name: northern assay evidence used in manual assertion
  'ECO:0005660':
    name: quantitative polymerase chain reaction evidence used in manual assertion
  'ECO:0001232':
    name: microscopy evidence used in manual assertion
  'ECO:0007087':
    name: ribosome profiling evidence
  'ECO:0007464':
    name: tryptic phosphopeptide mapping assay evidence used in automatic assertion
  'ECO:0007183':
    name: protein expression level evidence based on western blot used in manual assertion
  'ECO:0000295':
    name: RNA-sequencing evidence
  'ECO:0001838':
    name: in situ hybridization evidence used in manual assertion

suppressible_evidence_codes:

# the name of the db to use when creating a cvterm
db_name_for_cv: PBO

phenotype_cv_name: fission_yeast_phenotype
parental_strain:
  4897: 'yFS275'

# for these terms, if there is no evidence, use the given evidence code
auto_evidence_assignment:
  inviable vegetative cell population: Microscopy
  viable vegetative cell population: Microscopy
  elongated vegetative cell: Microscopy
  short mitotic spindle: Microscopy
  T-shaped cell: Microscopy
  pseudohyphal growth abolished: Microscopy
  abnormal mitotic spindle: Microscopy
  branched vegetative cell: Microscopy
  curved vegetative cell: Microscopy
  pear-shaped cell: Microscopy
  spheroid vegetative cell: Microscopy
  small cell: Microscopy
  stubby cell: Microscopy
  spheroid cell: Microscopy
  swollen cell: Microscopy
  abnormal microtubule cytoskeleton organization: Microscopy
  abnormal sporulation: Microscopy
  increased flocculation: Microscopy
  large and small daughter nuclei: Microscopy
  monopolar mitotic spindle: Microscopy
  long microtubules: Microscopy
  multinucleate vegetative cell: Microscopy
  elongated mutiseptate cell: Microscopy
  abnormal nuclear morphology during vegetative growth: Microscopy
  abnormal filament morphology: Microscopy
  abolished septum formation: Microscopy
  cut with abnormal chromosome segregation: Microscopy
  abnormal spindle assembly: Microscopy
  bottle-shaped cell: Microscopy
  shmoo with long tip: Microscopy
  mitochondria fused: Microscopy
  small vacuoles present in greater numbers during vegetative growth: Microscopy
  abnormal septation: Microscopy
  inviable elongated vegetative cell: Microscopy
  abnormal cell morphology: Microscopy
  lemon-shaped cell: Microscopy
  elongated multiseptate cell: Microscopy
  snowman-shaped spore: Microscopy
  lagging mitotic chromosomes: Microscopy
  mononucleate vegetative cell: Microscopy
  contractile ring absent: Microscopy
  inviable elongated dumbbell-shaped cell: Microscopy
  large vacuoles during vegetative growth: Microscopy
  abnormal kinetochore morphology: Microscopy
  abnormal cytokinesis: Microscopy
  resistance to 2-deoxyglucose: cell growth assay evidence
  resistance to 5-azacytidine: cell growth assay evidence
  resistance to brefeldin A: cell growth assay evidence
  resistance to thiabendazole: cell growth assay evidence
  resistance to amphotericin B: cell growth assay evidence
  resistance to aureobasidin A: cell growth assay evidence
  resistance to benomyl: cell growth assay evidence
  resistance to lovastatin: cell growth assay evidence
  resistance to nystatin: cell growth assay evidence
  resistance to rapamycin: cell growth assay evidence
  sensitive to enfumafungin: cell growth assay evidence
  sensitive to 6-azauracil: cell growth assay evidence
  sensitive to camptothecin: cell growth assay evidence
  sensitive to tacrolimus: cell growth assay evidence
  sensitive to hydrogen peroxide: cell growth assay evidence
  sensitive to hydroxyurea: cell growth assay evidence
  sensitive to methyl methanesulfonate: cell growth assay evidence
  sensitive to cadmium: cell growth assay evidence
  sensitive to caffeine: cell growth assay evidence
  sensitive to thiabendazole: cell growth assay evidence
  sensitive to cycloheximide: cell growth assay evidence
  sensitive to latrunculin A: cell growth assay evidence
  sensitive to rapamycin: cell growth assay evidence
  sensitive to valproic acid: cell growth assay evidence
  sensitive to zinc: cell growth assay evidence
  sensitive to heat: cell growth assay evidence
  sensitive to cold: cell growth assay evidence
  sensitive to ionizing radiation: cell growth assay evidence
  sensitive to UV during vegetative growth: cell growth assay evidence
  sensitive to cisplatin: cell growth assay evidence
  sensitive to high osmolarity: cell growth assay evidence
  sensitive to calcium: cell growth assay evidence
  sensitive to bleomycin: cell growth assay evidence
  sensitive to benomyl: cell growth assay evidence
  sensitive to salt stress: cell growth assay evidence
  sensitive to arsenic: cell growth assay evidence
  sensitive to N-ethylmaleimide: cell growth assay evidence
  sensitive to DNA damage: cell growth assay evidence
  sensitive to pravastatin: cell growth assay evidence
  decreased cell population growth: cell growth assay evidence
  abnormal cell population growth: cell growth assay evidence
  sensitive to menadione: cell growth assay evidence
  viable cell at high temperature: cell growth assay evidence
  resistance to L-azetidine-2-carboxylic acid: cell growth assay evidence
  resistance to K-252a: cell growth assay evidence
  normal cell population growth: cell growth assay evidence
  abnormal G1/S phase transcription: transcript expression level evidence
  abnormal actin cortical patch localization during vegetative growth: Microscopy
  abnormal actomyosin contractile ring myosin filament organization: Microscopy
  abnormal cellular respiration: cell growth assay evidence
  abnormal chromosome condensation: Microscopy
  abnormal chromosome segregation: Microscopy
  abnormal cytoplasmic microtubule nucleation during mitotic interphase: Microscopy
  abnormal endocytosis during vegetative growth: Microscopy
  abnormal glucan synthesis: Substance quantification evidence
  abnormal meiosis: Microscopy
  abnormal meiotic recombination: reporter gene assay evidence
  abnormal microtubule cytoskeleton organization during vegetative growth: Microscopy
  abnormal mitotic DNA damage checkpoint: Flow cytometry data
  abnormal mitotic sister chromatid segregation: Microscopy
  abnormal mitotic spindle assembly: Microscopy
  abnormal mitotic spindle elongation: Microscopy
  abnormal negative regulation of mitotic DNA replication initiation: Flow cytometry data
  abnormal plasma membrane sterol distribution: Microscopy
  abnormal regulation of translation in response to nitrogen starvation: cell growth assay evidence
  abnormal regulation of translation in response to osmotic stress: cell growth assay evidence
  abnormal regulation of translation in response to oxidative stress: cell growth assay evidence
  abnormal vegetative cell morphology: Microscopy
  abnormal vegetative cell polarity: Microscopy
  abolished DNA damage checkpoint override in response to caffeine: Flow cytometry data
  abolished cell-substrate adhesion: Other
  abolished crossover: reporter gene assay evidence
  actomyosin contractile ring absent: Microscopy
  "branched, elongated cell": Microscopy
  chromosome loss during mitotic chromosome segregation: reporter gene assay evidence
  decreased cell population growth at high temperature: cell growth assay evidence
  decreased cell population growth at low temperature: cell growth assay evidence
  decreased cell population growth on ammonia nitrogen source: cell growth assay evidence
  decreased cell population growth on galactose carbon source: cell growth assay evidence
  decreased cell population growth on proline nitrogen source: cell growth assay evidence
  decreased gene conversion during vegetative growth: reporter gene assay evidence
  decreased vegetative cell population growth: cell growth assay evidence
  elongated multiseptate vegetative cell: Microscopy
  fragmented DNA: Microscopy
  increased DNA recombination at mitotic DNA replication fork barriers: reporter gene assay evidence
  increased HMG-CoA reductase activity: Enzyme assay data
  increased centromeric outer repeat transcript level: transcript expression level evidence
  increased duration of mitotic M phase: Flow cytometry data
  increased long tract gene conversion: reporter gene assay evidence
  increased protein oxidation during vegetative growth: Western blot assay
  increased transcription from MCB promoter: transcript expression level evidence
  inviable elongated vegetative cell with central constriction: Microscopy
  mitochondrial nucleoid decreased number: Microscopy
  mitochondrial nucleoid enlarged: Microscopy
  mitotic G2/M transition delay: Microscopy
  mitotic chromosome fragmentation upon segregation: Microscopy
  mutator: reporter gene assay evidence
  normal growth on ammonia nitrogen source: cell growth assay evidence
  normal growth on proline nitrogen source: cell growth assay evidence
  normal vegetative cell population growth: cell growth assay evidence
  pear-shaped vegetative cell: Microscopy
  protein mislocalized to cytoplasm during vegetative growth: Microscopy
  sensitive to caffeine during vegetative growth: cell growth assay evidence
  sensitive to osmotic stress: cell growth assay evidence
  slow vegetative cell population growth: cell growth assay evidence
  small vegetative cell: Microscopy
  split actomyosin contractile ring: Microscopy
  sterile: cell growth assay evidence
  stubby vegetative cell: Microscopy
  swollen vegetative cell: Microscopy
  abnormal septum assembly: Microscopy

check_chado:
  'PomBase::Check::FeatureCount':
    min_count: 10000
  query_checks:
    - name: pombe_genes
      expected: '>= 6960'
      # if true, show the result set content on failure:
      verbose_fail: false
      query: |
        select * from feature f, cvterm cvt, organism o where
        f.type_id = cvt.cvterm_id and f.organism_id = o.organism_id and
        o.common_name = 'pombe' and cvt.name = 'gene'
    - name: enough_isa_cvterm_rels
      expected: '> 100000'
      verbose_fail: false
      query: |
        select * from cvterm_relationship
        where type_id in (select cvterm_id from cvterm where name = 'is_a')
    - name: no_duplicate_pombe_gene_names
      expected: 0
      # if true, show the result set content on failure:
      verbose_fail: true
      query: |
        select count(feature_id), name from feature where
        name is not null and type_id = (select cvterm_id from cvterm where cv_id
        = (select cv_id from cv where name = 'sequence') and name = 'gene') and
        name <> 'mat3-Mm' and name <> 'mat3-Mc' and
        organism_id = (select organism_id from organism where abbreviation =
        'Spombe') group by name having count(feature_id) > 1
    - name: alleles_instance_of_gene
      description: all alleles should be instance_of gene
      expected: 0
      verbose_fail: true
      query: |
        SELECT f.uniquename, f.name, pt.name from feature f
          JOIN feature_relationship r ON r.subject_id = f.feature_id
          JOIN feature p ON r.object_id = p.feature_id
          JOIN cvterm rt ON rt.cvterm_id = r.type_id
          JOIN cvterm pt ON pt.cvterm_id = p.type_id
          WHERE rt.name = 'instance_of' AND pt.name <> 'gene'
            AND pt.name <> 'pseudogene'
    - name: no_population_terms_with_penetrance
      description: no population term should have a has_penetrance relation
      expected: 0
      verbose_fail: true
      query: |
        SELECT f.name, f.uniquename, res_fc.cvterm_name, pub.uniquename
          FROM pombase_feature_cvterm_ext_resolved_terms res_fc
               JOIN pub on res_fc.pub_id = pub.pub_id
               JOIN feature f on f.feature_id = res_fc.feature_id
               JOIN cvtermpath path on path.subject_id = res_fc.base_cvterm_id
          WHERE object_id in
            (SELECT cvterm_id
               FROM cvterm t join cv on t.cv_id = cv.cv_id
              WHERE t.name = 'cell population phenotype' and cv.name = 'fission_yeast_phenotype')
            AND pathdistance > 0
            AND cvterm_id in
              (SELECT subject_id
                 FROM cvterm_relationship r
                      JOIN cvterm type on r.type_id = type.cvterm_id
                      AND type.name = 'has_penetrance')
            AND path.type_id in (select cvterm_id from cvterm where name = 'is_a')
    - name: canto_annotations
      description: check for annotations from Canto
      expected: '>= 280'
      verbose_fail: false
      query: |
        SELECT fc.feature_cvterm_id
          FROM feature_cvterm fc
          JOIN feature_cvtermprop p ON p.feature_cvterm_id = fc.feature_cvterm_id
          JOIN cvterm type ON p.type_id = type.cvterm_id
         WHERE type.name = 'canto_session'
    - name: annotation_count
      description: check that we have enough annotations
      expected: '>= 90000'
      verbose_fail: false
      query: |
        SELECT fc.feature_cvterm_id
          FROM feature_cvterm fc
    - name: go_annotation_count
      description: "check that GO filtering hasn't failed"
      expected: '<= 40000'
      verbose_fail: false
      query: |
        SELECT fc.feature_cvterm_id
          FROM feature_cvterm fc
          JOIN cvterm t ON t.cvterm_id = fc.cvterm_id
          JOIN cv ON t.cv_id = cv.cv_id
        WHERE cv.name IN ('biological_process',
                          'molecular_function',
                          'cellular_component')
    - name: quantitative_annotation_qualifiers
      description: check that all quantitative annotation have a count qualifier
      expected: 0
      verbose_fail: false
      query: |
        SELECT feature_cvterm_id
          FROM pombase_feature_cvterm_ext_resolved_terms fc
          JOIN feature f ON f.feature_id = fc.feature_id
         WHERE feature_cvterm_id NOT IN
           (SELECT feature_cvterm_id
              FROM feature_cvtermprop
             WHERE type_id IN
               (SELECT cvterm_id
                  FROM cvterm
                 WHERE name = 'quant_gene_ex_avg_copies_per_cell'))
           AND (base_cvterm_name = 'protein level' OR base_cvterm_name = 'RNA level')
           AND base_cvterm_name = 'protein level'
           AND feature_cvterm_id NOT IN
             (SELECT feature_cvterm_id
                FROM feature_cvtermprop
               WHERE type_id IN (SELECT cvterm_id FROM cvterm WHERE name = 'qualifier'))
    - name: missing_assayed_using
      description: check that protein binding annotations have two assayed_using extensions
      expected: 0
      verbose_fail: true
      query: |
        SELECT DISTINCT fc.feature_cvterm_id, f.uniquename || ' <-> ' || cvterm_name,
                 pub.uniquename,
                 (SELECT value
                    FROM feature_cvtermprop p
                   WHERE p.feature_cvterm_id = fc.feature_cvterm_id
                     AND p.type_id IN (SELECT cvterm_id FROM cvterm WHERE name = 'canto_session') LIMIT 1) AS session
          FROM pombase_feature_cvterm_ext_resolved_terms t
          JOIN feature_cvterm fc ON fc.cvterm_id = t.cvterm_id
          JOIN feature f ON f.feature_id = fc.feature_id
          JOIN pub ON fc.pub_id = pub.pub_id
         WHERE
           base_cvterm_id IN
             (SELECT t.cvterm_id
                FROM cvterm t JOIN dbxref x ON t.dbxref_id = x.dbxref_id
                              JOIN db ON db.db_id = x.db_id
               WHERE db.name = 'FYPO'
                 AND x.accession IN
                   ('0000702', '0000703', '0000704', '0000705', '0001275', '0001571', '0001645', '0002365', '0003206', '0003207', '0003215', '0004335'))
           AND t.cvterm_id IN
             (SELECT cvterm_id FROM pombase_extension_rels_and_values
               WHERE rel_name = 'assayed_using' GROUP BY cvterm_id, rel_name HAVING COUNT(value) = 1)
    - name: no_duplicate_orthologs
      description: "don't store the same ortholog twice"
      expected: 0
      verbose_fail: true
      query: |
        SELECT
          tf.uniquename, ef.uniquename, count(*)
        FROM
          feature tf
          JOIN feature_relationship fr ON (tf.feature_id=fr.object_id)
          JOIN cvterm frt ON (fr.type_id=frt.cvterm_id)
          JOIN feature ef ON (fr.subject_id=ef.feature_id)
        WHERE
          frt.name='orthologous_to'
        GROUP BY tf.uniquename, ef.uniquename HAVING COUNT(*)>1
    - name: enough_genetic_interactions
      expected: '> 6'
      query: |
        SELECT * FROM feature_relationship fr JOIN cvterm frt ON (fr.type_id = frt.cvterm_id)
          WHERE frt.name = 'interacts_genetically'
    - name: enough_physical_interactions
      expected: '> 14'
      query: |
        SELECT * FROM feature_relationship fr JOIN cvterm frt ON (fr.type_id = frt.cvterm_id)
          WHERE frt.name = 'interacts_physically'
    - name: only_ascii_characters_in_feature_name
      description: Check for non-ascii characters in feature names
      expected: 0
      verbose_fail: true
      query: |
        SELECT uniquename,
               name,
          (SELECT distinct value
           FROM featureprop p
           JOIN cvterm pt ON pt.cvterm_id = p.type_id
           WHERE p.feature_id = f.feature_id
             AND pt.name = 'canto_session'
           LIMIT 1) AS session
        FROM feature f
        WHERE length(name) <> length(regexp_replace(name, '[^\x01-\x7f]', '', 'g'))
    - name: extension_relation_genes_exist
      description: Check that gene identifiers used in extensions exist
      expected: 0
      verbose_fail: true
      query: |
        SELECT value FROM cvtermprop
          WHERE type_id IN
            (SELECT cvterm_id FROM cvterm
              WHERE name LIKE 'annotation_extension_relation-%')
                AND value LIKE 'SP%'
                AND value NOT IN (SELECT uniquename FROM feature WHERE uniquename LIKE 'SP%')
    - name: annotation_with_no_evidence
      verbose_fail: true
      expected: 0
      query: |
          SELECT pub.uniquename AS pmid, t.name AS term_name, allele.name, allele.uniquename
          FROM feature_cvterm fc
          JOIN pub ON fc.pub_id = pub.pub_id
          JOIN feature f ON f.feature_id = fc.feature_id
          JOIN cvterm t ON fc.cvterm_id = t.cvterm_id
          JOIN cv ON t.cv_id = cv.cv_id
          LEFT OUTER JOIN feature_relationship feat_rel ON fc.feature_id = feat_rel.object_id
          LEFT OUTER JOIN feature allele ON feat_rel.subject_id = allele.feature_id
          WHERE cv.name IN ('molecular_function', 'cellular_component', 'biological_process', 'fission_yeast_phenotype')
            AND fc.feature_cvterm_id NOT IN
              (SELECT feature_cvterm_id FROM feature_cvtermprop p
               WHERE p.type_id IN (SELECT cvterm_id FROM cvterm WHERE name = 'evidence'))
            AND feat_rel.type_id IN
              (SELECT cvterm_id FROM cvterm WHERE name = 'part_of')
    - name: duplicate_go_annotation
      verbose_fail: true
      description: "All duplicate GO annotation.  Columns are: count, systematic ID, gene name, term name, evidence, PubMed ID, 'with', session ID"
      expected: 0
      query: |
         SELECT count(fc.feature_cvterm_id), f.uniquename as systematic_id, gene.name as gene_name, cvterm_name, evprop.value as evidence, pub.uniquename as pmid, withp.value as with_prop,
                  array_to_string(array(select value from pubprop where pubprop.pub_id = pub.pub_id and pubprop.type_id in (select cvterm_id from cvterm where name = 'canto_session') ORDER BY value), ',') as session
           FROM feature f
           JOIN feature_relationship rel ON rel.subject_id = f.feature_id
           JOIN feature gene on rel.object_id = gene.feature_id
           JOIN pombase_feature_cvterm_ext_resolved_terms fc ON fc.feature_id = f.feature_id
           JOIN pub ON fc.pub_id = pub.pub_id
           LEFT OUTER JOIN feature_cvtermprop evprop ON evprop.feature_cvterm_id = fc.feature_cvterm_id and evprop.type_id in (select cvterm_id from cvterm where name = 'evidence')
           LEFT OUTER JOIN feature_cvtermprop afp ON afp.feature_cvterm_id = fc.feature_cvterm_id and afp.type_id = (select cvterm_id from cvterm where name = 'gene_product_form_id')
           LEFT OUTER JOIN feature_cvtermprop withp ON withp.feature_cvterm_id = fc.feature_cvterm_id and withp.type_id = (select cvterm_id from cvterm where name = 'with')
          WHERE (fc.base_cv_name = 'cellular_component' OR fc.base_cv_name = 'biological_process' OR fc.base_cv_name = 'molecular_function')
          GROUP BY f.uniquename, gene.name, cvterm_name, evidence, pub.uniquename, afp.value, with_prop, session
         HAVING count(fc.feature_cvterm_id) > 1;
    - name: illegal_with_prefix
      verbose_fail: true
      description: "'with' values that have an unsupported DB prefix"
      expected: 0
      query: |
         SELECT distinct substring(value FROM '(.*):.*') AS with_db_prefix,
            (SELECT value FROM feature_cvtermprop p2
             JOIN cvterm p2t ON p2.type_id = p2t.cvterm_id
             WHERE p2t.name = 'canto_session' AND p.feature_cvterm_id = p2.feature_cvterm_id) AS SESSION,
                 pub.uniquename AS pubmed_id
          FROM feature_cvtermprop p
          JOIN cvterm t ON p.type_id = t.cvterm_id
          JOIN feature_cvterm fc ON fc.feature_cvterm_id = p.feature_cvterm_id
          JOIN pub ON pub.pub_id = fc.pub_id
          WHERE t.name = 'with'
            AND substring(value FROM '(.*):.*')
                NOT IN ('AGI_LocusCode', 'CBS', 'CGD', 'dictyBase', 'EC', 'ECOGENE', 'FB', 'GO',
                        'HGNC', 'InterPro', 'MGI', 'PANTHER', 'PDB', 'Pfam', 'PomBase', 'RGD',
                        'SGD', 'SUPERFAMILY', 'TAIR', 'TAIR:locus', 'UniPathway', 'UniProtKB', 'UniProtKB-KW',
                        'UniProtKB-SubCell', 'UniRule', 'WB', 'ZFIN', 'JaponicusDB');
    - name: prop_values_with_missing_terms
      verbose_fail: true
      description: "relation ranges / annotation properties where the term is missing from the database, eg. from(...) or column_17(...)"
      github_issue: "https://github.com/pombase/pombase-chado/issues/693"
      expected: 0
      query: |
         WITH all_term_ids AS
           (SELECT db.name || ':' || x.accession AS term_id
            FROM dbxref AS x
            JOIN db ON db.db_id = x.db_id
            JOIN cvterm ON cvterm.dbxref_id = x.dbxref_id
            WHERE db.name IN ('GO', 'SO', 'PR'))
         SELECT
           (SELECT value
            FROM feature_cvtermprop p
            WHERE p.feature_cvterm_id = fc.feature_cvterm_id
              AND type_id IN
                (SELECT cvterm_id
                 FROM cvterm
                 WHERE name = 'canto_session')),
           f.uniquename, t.name, fcpt.name, fcp.value
         FROM feature_cvtermprop fcp
         JOIN cvterm fcpt ON fcp.type_id = fcpt.cvterm_id
         JOIN feature_cvterm fc ON fcp.feature_cvterm_id = fc.feature_cvterm_id
         JOIN feature f ON fc.feature_id = f.feature_id
         JOIN cvterm t ON fc.cvterm_id = t.cvterm_id
         WHERE fcpt.name in ('gene_product_form_id', 'from', 'with')
           AND fcp.value ~ '^(GO|SO|PR):\d+$'
           AND fcp.value NOT IN
             (SELECT term_id
              FROM all_term_ids);
    - name: duplicate_allele_descriptions
      verbose_fail: true
      description: "alleles with the same description that have different names"
      github_issue: "https://github.com/pombase/canto/issues/828"
      expected: 0
      query: |
         SELECT gene.uniquename,
                allele1.name as allele1_name,
                (select value from featureprop fp where fp.feature_id = allele1.feature_id AND fp.type_id in (select cvterm_id from cvterm where name = 'canto_session')  limit 1) as allele1_session,
                allele1_prop.value AS description,
                allele2.name as allele2_name,
                (select value from featureprop fp where fp.feature_id = allele2.feature_id AND fp.type_id in (select cvterm_id from cvterm where name = 'canto_session')  limit 1) as allele2_session
         FROM feature allele1
         JOIN featureprop allele1_prop ON allele1_prop.feature_id = allele1.feature_id
         JOIN cvterm a1_prop_type ON a1_prop_type.cvterm_id = allele1_prop.type_id
         JOIN feature_relationship rel1 ON allele1.feature_id = rel1.subject_id
         JOIN cvterm rel1_type ON rel1_type.cvterm_id = rel1.type_id
         JOIN feature gene ON rel1.object_id = gene.feature_id
         JOIN feature_relationship rel2 ON rel2.object_id = gene.feature_id
         JOIN cvterm rel2_type ON rel2.type_id = rel2_type.cvterm_id
         JOIN feature allele2 ON rel2.subject_id = allele2.feature_id
         JOIN featureprop allele2_prop ON allele2_prop.feature_id = allele2.feature_id
         JOIN cvterm a2_prop_type ON a2_prop_type.cvterm_id = allele2_prop.type_id
         WHERE allele1.feature_id <> allele2.feature_id
           AND allele1.name <> allele2.name
           AND rel1_type.name = 'instance_of'
           AND rel2_type.name = 'instance_of'
           AND a1_prop_type.name = 'description'
           AND a2_prop_type.name = 'description'
           AND allele1_prop.value = allele2_prop.value
           AND allele1_prop.value <> 'unknown';

external_sources:
  pubmed_efetch_url: 'http://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=pubmed&retmode=xml&retmax=99999&id='
  pubmed_esearch_url: 'http://eutils.ncbi.nlm.nih.gov/entrez/eutils/esearch.fcgi?db=pubmed&remode=xml&retmax=99999&term='