Trick: uri-encode problematic fields.
Signed-off-by: Munyoki Kilyungi <me@bonfacemunyoki.com>
1 files changed, 134 insertions, 0 deletions
diff --git a/examples/probesets.scm b/examples/probesets.scm
new file mode 100755
index 0000000..11b867a
--- /dev/null
+++ b/examples/probesets.scm
@@ -0,0 +1,134 @@
+#! /usr/bin/env guile
+!#
+
+(use-modules (srfi srfi-1)
+ (srfi srfi-26)
+ (ice-9 getopt-long)
+ (ice-9 match)
+ (ice-9 regex)
+ (transform strings)
+ (transform sql)
+ (transform triples)
+ (transform special-forms)
+ (web uri))
+
+(define-transformer probeset->metadata
+ (tables (ProbeSet
+ (left-join GeneChip "ON GeneChip.Id = ProbeSet.ChipId"))
+ "WHERE ProbeSet.Name IS NOT NULL AND TRIM(ProbeSet.Name) != ''")
+ (triples
+ (string->identifier "probeset" (field ProbeSet Name))
+ (set rdf:type 'gnc:probeset)
+ (set skos:prefLabel (field ProbeSet Name))
+ (multiset skos:altLabel
+ (map string-trim-both
+ (string-split (sanitize-rdf-string (field ProbeSet alias)) #\;)))
+ (set gnt:uses_genechip (string->identifier "platform" (field GeneChip Name) #:separator "_"))
+ (set gnt:has_target_id (string-trim-both (sanitize-rdf-string (field ProbeSet TargetId))))
+ (set gnt:symbol (string-trim-both (field ProbeSet Symbol)))
+ (set dct:description (sanitize-rdf-string (field ProbeSet description)))
+ (set gnt:targets_region (string-trim-both (sanitize-rdf-string (field ProbeSet Probe_set_target_region))))
+ (set gnt:chr (field ProbeSet Chr))
+ (set gnt:mb (annotate-field (field ("IFNULL(ProbeSet.Mb, '')" Mb)) '^^xsd:double))
+ (set gnt:mb_mm8 (annotate-field (field ("IFNULL(ProbeSet.Mb_mm8, '')" Mb_mm8))
+ '^^xsd:double))
+ (set gnt:has_specificity
+ (field ("IFNULL(ProbeSet.Probe_set_specificity, '')"
+ Probe_set_specificity)))
+ (set gnt:has_blat_score
+ (field ("IFNULL(ProbeSet.Probe_set_BLAT_score, '')"
+ Probe_set_BLAT_score)))
+ (set gnt:has_blat_mb_start
+ (annotate-field (field ("IFNULL(ProbeSet.Probe_set_Blat_Mb_start, '')"
+ Probe_set_Blat_Mb_start))
+ '^^xsd:double))
+ (set gnt:has_blat_mb_end
+ (annotate-field (field ("IFNULL(ProbeSet.Probe_set_Blat_Mb_end, '')"
+ Probe_set_Blat_Mb_end))
+ '^^xsd:double))
+ (set gnt:has_blat_seq (sanitize-rdf-string (field ProbeSet BlatSeq)))
+ (set gnt:has_target_seq (sanitize-rdf-string (field ProbeSet TargetSeq)))
+ (set gnt:has_homologene_id (ontology 'homologene:
+ (uri-encode
+ (field ("IFNULL(ProbeSet.HomoloGeneID, '')"
+ HomoloGeneID)))))
+ (set gnt:has_uniprot_id (ontology 'uniprot:
+ (uri-encode
+ (field ("IFNULL(ProbeSet.UniProtID, '')"
+ UniProtID)))))
+ (set gnt:has_pub_chem_id (ontology
+ 'pubchem:
+ (uri-encode
+ (field ("IFNULL(ProbeSet.PubChem_ID, '')"
+ PubChem_ID)))))
+ (set gnt:has_kegg_id (ontology
+ 'kegg:
+ (uri-encode
+ (field ("IFNULL(ProbeSet.KEGG_ID, '')"
+ KEGG_ID)))))
+ (set gnt:has_omim_id (ontology
+ 'omim:
+ (uri-encode
+ (let ((omim (field ("IFNULL(ProbeSet.OMIM, '')"
+ OMIM))))
+ (if (number? omim)
+ omim
+ (regexp-substitute/global
+ #f "[^0-9]"
+ omim
+ 'pre "" 'post))))))
+ (set gnt:has_chebi_id (ontology
+ 'chebi:
+ (uri-encode
+ (field ("IFNULL(ProbeSet.ChEBI_ID, '')"
+ ChEBI_ID)))))))
+
+
+
+
+(let* ((option-spec
+ '((settings (single-char #\s) (value #t))
+ (output (single-char #\o) (value #t))
+ (documentation (single-char #\d) (value #t))))
+ (options (getopt-long (command-line) option-spec))
+ (settings (option-ref options 'settings #f))
+ (output (option-ref options 'output #f))
+ (documentation (option-ref options 'documentation #f))
+ (%connection-settings
+ (call-with-input-file settings
+ read)))
+ (call-with-target-database
+ %connection-settings
+ (lambda (db)
+ (with-documentation
+ (name "ProbeSet Metadata")
+ (connection %connection-settings)
+ (table-metadata? #f)
+ (total-rows (assoc-ref
+ (sql-find db "SELECT count(*) AS count from ProbeSet")
+ "count"))
+ (rows-per-chunk 1000000)
+ (prefixes
+ '(("gn:" "<http://genenetwork.org/id/>")
+ ("probeset:" "<http://genenetwork.org/probeset/>")
+ ("gnc:" "<http://genenetwork.org/category/>")
+ ("gnt:" "<http://genenetwork.org/term/>")
+ ("rdf:" "<http://www.w3.org/1999/02/22-rdf-syntax-ns#>")
+ ("kegg:" "<http://bio2rdf.org/ns/kegg#>")
+ ("pubchem:" "<https://pubchem.ncbi.nlm.nih.gov/>")
+ ("omim:" "<https://www.omim.org/entry/>")
+ ("rdfs:" "<http://www.w3.org/2000/01/rdf-schema#>")
+ ("uniprot:" "<http://purl.uniprot.org/uniprot/>")
+ ("chebi:" "<http://purl.obolibrary.org/obo/CHEBI_>")
+ ("dct:" "<http://purl.org/dc/terms/>")
+ ("owl:" "<http://www.w3.org/2002/07/owl#>")
+ ("homologene:" "<https://bio2rdf.org/homologene:>")
+ ("xsd:" "<http://www.w3.org/2001/XMLSchema#>")
+ ("qb:" "<http://purl.org/linked-data/cube#>")
+ ("sdmx-measure:" "<http://purl.org/linked-data/sdmx/2009/measure#>")
+ ("skos:" "<http://www.w3.org/2004/02/skos/core#>")))
+ (inputs
+ (list probeset->metadata))
+ (outputs
+ `(#:documentation ,documentation
+ #:rdf ,output))))))
|
