def rdf_level(gene_id, condition_label, base_condition_label,
ext_base_condition_label, fold_change, pvalue, time_point):
# TODO: it would be more correct to not change the case, but we want Knet compatibility
exp_uri = make_ae_exp_uri(exp_acc)
gene_id_nrm = gene_id.lower()
gene_uri = make_gene_uri(gene_id)
cond_id = make_id(condition_label, skip_non_word_chars=False)
cond_uri = make_condition_uri(condition_label)
base_cond_id = make_id(base_condition_label, skip_non_word_chars=False)
base_cond_uri = make_condition_uri(base_condition_label)
ext_base_cond_id, ext_base_cond_uri = None, None
if ext_base_condition_label:
ext_base_cond_id = make_id(ext_base_condition_label,
skip_non_word_chars=False)
ext_base_cond_uri = make_condition_uri(ext_base_condition_label)
exp_stmt_uri = f"bkr:gxaexp_{exp_acc}_{gene_id_nrm}_{cond_id}_vs_{base_cond_id}"
if ext_base_cond_uri:
exp_stmt_uri += f"_in_{ext_base_cond_id}"
if time_point != -1: exp_stmt_uri += f"_{time_point}h"
rdf = f"""
{exp_stmt_uri} a rdfs:Statement;
rdf:subject {gene_uri};
rdf:predicate bioschema:expressedIn;
rdf:object {cond_uri};
agri:baseCondition {base_cond_uri};
agri:log2FoldChange {fold_change};
agri:pvalue {pvalue};
agri:evidence {exp_uri}
.
{gene_uri} bioschema:expressedIn {cond_uri}.
"""
rdf = dedent(rdf)
if ext_base_cond_uri:
rdf += f"\n{exp_stmt_uri} agri:baseCondition {ext_base_cond_uri}.\n"
if time_point != -1:
time_point_str = str(time_point) + " hours"
time_point_uri = make_condition_uri(time_point_str)
rdf += f"\n{exp_stmt_uri} agri:timePoint {time_point_uri}."
print(rdf, file=out)
def rdf_specie(exp_uri: str, specie_label: str):
if not specie_label: return ""
specie_uri = "bkr:specie_" + make_id(specie_label,
skip_non_word_chars=True)
rdf = f"""
{exp_uri} schema:additionalProperty {specie_uri}.
{specie_uri} a schema:PropertyValue;
schema:propertyID "organism";
schema:value "{specie_label}";
"""
rdf = dedent(rdf)
specie_terms = specie2terms.get(specie_label)
if specie_terms:
rdf_terms = ", ".join(["<" + s + ">" for s in specie_terms])
rdf += "\tdc:type " + rdf_terms + ";\n"
rdf += ".\n"
return rdf
def rdf_level(gene_id, condition_label, tpm, ordinal_tpm):
# TODO: it would be more correct to not change the case, but we want Knet compatibility
exp_uri = make_ae_exp_uri(exp_acc)
gene_id_nrm = gene_id.lower()
gene_uri = make_gene_uri(gene_id)
cond_id = make_id(condition_label, skip_non_word_chars=False)
cond_uri = make_condition_uri(condition_label)
rdf = f"""
bkr:gxaexp_{exp_acc}_{gene_id_nrm}_{cond_id} a rdfs:Statement;
agri:tpmCount {tpm};
agri:ordinalTpm "{ordinal_tpm}";
rdf:subject {gene_uri};
rdf:predicate bioschema:expressedIn;
rdf:object {cond_uri};
agri:evidence {exp_uri}
.
{gene_uri} bioschema:expressedIn {cond_uri}.
"""
rdf = dedent(rdf)
print(rdf, file=out)
def rdf_ae_experiment(exp_js: dict, out=stdout) -> str:
specie2terms = {
"arabidopsis thaliana":
["http://purl.bioontology.org/ontology/NCBITAXON/3701"],
"triticum aestivum":
["http://purl.bioontology.org/ontology/NCBITAXON/4565"]
}
def rdf_specie(exp_uri: str, specie_label: str):
if not specie_label: return ""
specie_uri = "bkr:specie_" + make_id(specie_label,
skip_non_word_chars=True)
rdf = f"""
{exp_uri} schema:additionalProperty {specie_uri}.
{specie_uri} a schema:PropertyValue;
schema:propertyID "organism";
schema:value "{specie_label}";
"""
rdf = dedent(rdf)
specie_terms = specie2terms.get(specie_label)
if specie_terms:
rdf_terms = ", ".join(["<" + s + ">" for s in specie_terms])
rdf += "\tdc:type " + rdf_terms + ";\n"
rdf += ".\n"
return rdf
def rdf_publication(exp_uri, exp_js):
if "bibliography" not in exp_js: return ""
rdf = ""
for pub_js in exp_js["bibliography"]:
# Without this very minimum, it's hardly a meaningful entry
if not ("title" in pub_js or "accession" in pub_js
or "doi" in pub_js):
continue
if "accession" in pub_js:
pub_uri = "bkr:pmid_" + str(pub_js["accession"])
elif "doi" in pub_js:
pub_uri = pub_js["doi"]
else:
pub_uri = "bkr:pub_" + hash_generator(pub_js.values())
rdf += f"""
{exp_uri} schema:subjectOf {pub_uri}.
{pub_uri} a agri:ScholarlyPublication;
"""
rdf = dedent(rdf)
rdf += rdf_text(pub_js, "title", "\tdc:title")
rdf += rdf_text(pub_js, "authors", "\tagri:authorsList")
rdf += rdf_str(pub_js, "accession", "\tagri:pmedId")
rdf += rdf_str(pub_js, "doi", "\tagri:doiId")
rdf += rdf_str(pub_js, "year", "\tschema:datePublished")
rdf += ".\n"
return dedent(rdf)
exp_acc = exp_js["accession"]
exp_uri = make_ae_exp_uri(exp_acc)
rdf = f"""
{exp_uri} a bioschema:Study;
schema:identifier "{exp_acc}";
"""
rdf = dedent(rdf)
rdf += rdf_text(exp_js, "name", "\tdc:title")
# TODO: not clear why they're arrays
if exp_js["description"]:
rdf += rdf_text(exp_js["description"][0], "text",
"\tschema:description")
rdf += rdf_str(exp_js, "releasedate", "\tschema:datePublished")
# gxaAnalysisType is added by gxa.gxa_get_experiment_descriptors() and they can be 'Differential', 'Baseline'
# Detailed specifications for such types are in gxa-defaults.ttl, here we create a link to the corresponding
# URIs used there
#
rdf += rdf_pval(
exp_js, "gxaAnalysisType", "\tschema:additionalProperty",
lambda gxa_type: "bkr:gxa_analysis_type_" + make_id(
gxa_type, skip_non_word_chars=True))
rdf += ".\n"
for specie in exp_js.get("organism", []):
rdf += rdf_specie(exp_uri, specie)
rdf += rdf_publication(exp_uri, exp_js)
if out:
print(rdf, file=out)
else:
return rdf
def make_condition_uri(condition_label: str) -> str:
return "bkr:cond_" + make_id(condition_label, skip_non_word_chars=False)