COMPLoinc Example

See https://github.com/loinc/comp-loinc for a description of the COMPLoinc project. This creates an OWL version of LOINC that can be used in OAK to explore relationships between codes and components.

Currently this notebook largely only uses the command line functionality of OAK. It should therefore be accessible to non-programmers (although it helps to have a good understanding of the command line, and some advanced OAK query concepts are introduced).

In future we may extend this notebook to have Python examples

Creating an alias

First we create an alias comploinc for the runoak command using a sqlite selector for the comploinc resource

[1]:

%alias comploinc runoak -i sqlite:obo:comploinc

Basic Lookup

The info command can be used to do lookup for a given entity or set of entities. These can be specified as lists of CURIEs or labels on the command line after the info command:

[2]:

comploinc info loinc:11145-0

loinc:11145-0 ! 5-Hydroxyindoleacetate/Creatinine:MRto:Pt:Urine:Qn:

Graph Visualization

Here we show the viz command with a single term (multiple terms can be passed, but we illustrate with one for now)

[7]:

comploinc viz loinc:11145-0  -o output/loinc-11145-0.png

Lookup by label

[4]:

comploinc info Serotonin

loinc:LP14693-3 ! Serotonin

[5]:

comploinc descendants -p i Serotonin

loinc:LP14693-3 ! Serotonin
loinc:LP15097-6 ! 5-Hydroxyindoleacetate
loinc:LP36806-5 ! 5-Hydroxyindoleacetate & Creatinine

[6]:

comploinc descendants -p i,loinc:hasComponent Serotonin

loinc:11145-0 ! 5-Hydroxyindoleacetate/Creatinine:MRto:Pt:Urine:Qn:
loinc:12172-3 ! 5-Hydroxyindoleacetate:PrThr:24H:Urine:Ord:
loinc:14573-0 ! 5-Hydroxyindoleacetate:SRat:24H:Urine:Qn:
loinc:14910-4 ! Serotonin:SCnc:Pt:Ser:Qn:
loinc:15009-4 ! 5-Hydroxyindoleacetate:SCnc:Pt:Urine:Qn:
loinc:1692-3 ! 5-Hydroxyindoleacetate:MCnc:Pt:CSF:Qn:
loinc:1693-1 ! 5-Hydroxyindoleacetate:MCnc:Pt:Ser/Plas:Qn:
loinc:1694-9 ! 5-Hydroxyindoleacetate:MCnc:Pt:Urine:Qn:
loinc:1695-6 ! 5-Hydroxyindoleacetate:MRat:24H:Urine:Qn:
loinc:17003-5 ! Serotonin:MCnc:Pt:Urine:Qn:
loinc:18253-5 ! Serotonin:MRat:24H:Urine:Qn:
loinc:18375-6 ! Serotonin:MCnc:Pt:Urine:Qn:
loinc:25524-0 ! Serotonin:SCnc:Pt:Bld:Qn:
loinc:25971-3 ! 5-Hydroxyindoleacetate:SCnc:24H:Urine:Qn:
loinc:25981-2 ! Serotonin:SCnc:24H:Urine:Qn:
loinc:26035-6 ! Serotonin:SCnc:Pt:Plas:Qn:
loinc:27057-9 ! Serotonin:MCnc:Pt:Ser:Qn:
loinc:2939-7 ! Serotonin:MCnc:Pt:Bld:Qn:
loinc:2940-5 ! Serotonin:MCnc:Pt:Plas:Qn:
loinc:2941-3 ! Serotonin:MCnc:Pt:Platelets:Qn:
loinc:29520-4 ! 5-Hydroxyindoleacetate/Creatinine:SRto:Pt:Urine:Qn:
loinc:31203-3 ! 5-Hydroxyindoleacetate:MCnc:24H:Urine:Qn:
loinc:32339-4 ! Serotonin:SRat:24H:Urine:Qn:
loinc:34373-1 ! Serotonin:SCnc:Pt:Urine:Qn:
loinc:34374-9 ! Serotonin/Creatinine:SRto:Pt:Urine:Qn:
loinc:42671-8 ! Serotonin:EntMass:Pt:Platelets:Qn:
loinc:44288-9 ! 5-Hydroxyindoleacetate/Creatinine:MRto:24H:Urine:Qn:
loinc:44909-0 ! 5-Hydroxyindoleacetate & Creatinine:Imp:Pt:Urine:Nom:
loinc:47544-2 ! 5-Hydroxyindoleacetate:SCnc:Pt:CSF:Qn:
loinc:47545-9 ! 5-Hydroxyindoleacetate/Creatinine:SRto:24H:Urine:Qn:
loinc:48168-9 ! 5-Hydroxyindoleacetate:PrThr:Pt:Urine:Ord:
loinc:50149-4 ! 5-Hydroxyindoleacetate:SCnc:Pt:Ser/Plas:Qn:
loinc:56978-0 ! Serotonin:MCnc:24H:Urine:Qn:
loinc:71804-9 ! Serotonin:EntSub:Pt:Platelets:Qn:
loinc:74769-1 ! 5-Hydroxyindoleacetate:SCnc:Pt:PRP:Qn:
loinc:LP14693-3 ! Serotonin
loinc:LP15097-6 ! 5-Hydroxyindoleacetate
loinc:LP36806-5 ! 5-Hydroxyindoleacetate & Creatinine

Text Search

[7]:

comploinc info l~Serotonin

loinc:LP14693-3 ! Serotonin
loinc:14910-4 ! Serotonin:SCnc:Pt:Ser:Qn:
loinc:17003-5 ! Serotonin:MCnc:Pt:Urine:Qn:
loinc:18253-5 ! Serotonin:MRat:24H:Urine:Qn:
loinc:18375-6 ! Serotonin:MCnc:Pt:Urine:Qn:
loinc:25524-0 ! Serotonin:SCnc:Pt:Bld:Qn:
loinc:25981-2 ! Serotonin:SCnc:24H:Urine:Qn:
loinc:26035-6 ! Serotonin:SCnc:Pt:Plas:Qn:
loinc:27057-9 ! Serotonin:MCnc:Pt:Ser:Qn:
loinc:2939-7 ! Serotonin:MCnc:Pt:Bld:Qn:
loinc:2940-5 ! Serotonin:MCnc:Pt:Plas:Qn:
loinc:2941-3 ! Serotonin:MCnc:Pt:Platelets:Qn:
loinc:32339-4 ! Serotonin:SRat:24H:Urine:Qn:
loinc:34373-1 ! Serotonin:SCnc:Pt:Urine:Qn:
loinc:34374-9 ! Serotonin/Creatinine:SRto:Pt:Urine:Qn:
loinc:42671-8 ! Serotonin:EntMass:Pt:Platelets:Qn:
loinc:56978-0 ! Serotonin:MCnc:24H:Urine:Qn:
loinc:71804-9 ! Serotonin:EntSub:Pt:Platelets:Qn:

Boolean Graph Queries

Find all codes that have a component of “Serotonin” and has a system of “Urine”

[8]:

comploinc info .descendant//p=i,loinc:hasComponent Serotonin .and .descendant//p=i,loinc:hasSystem Urine

loinc:15009-4 ! 5-Hydroxyindoleacetate:SCnc:Pt:Urine:Qn:
loinc:1695-6 ! 5-Hydroxyindoleacetate:MRat:24H:Urine:Qn:
loinc:44909-0 ! 5-Hydroxyindoleacetate & Creatinine:Imp:Pt:Urine:Nom:
loinc:34373-1 ! Serotonin:SCnc:Pt:Urine:Qn:
loinc:25971-3 ! 5-Hydroxyindoleacetate:SCnc:24H:Urine:Qn:
loinc:31203-3 ! 5-Hydroxyindoleacetate:MCnc:24H:Urine:Qn:
loinc:18253-5 ! Serotonin:MRat:24H:Urine:Qn:
loinc:34374-9 ! Serotonin/Creatinine:SRto:Pt:Urine:Qn:
loinc:32339-4 ! Serotonin:SRat:24H:Urine:Qn:
loinc:29520-4 ! 5-Hydroxyindoleacetate/Creatinine:SRto:Pt:Urine:Qn:
loinc:17003-5 ! Serotonin:MCnc:Pt:Urine:Qn:
loinc:25981-2 ! Serotonin:SCnc:24H:Urine:Qn:
loinc:44288-9 ! 5-Hydroxyindoleacetate/Creatinine:MRto:24H:Urine:Qn:
loinc:12172-3 ! 5-Hydroxyindoleacetate:PrThr:24H:Urine:Ord:
loinc:56978-0 ! Serotonin:MCnc:24H:Urine:Qn:
loinc:18375-6 ! Serotonin:MCnc:Pt:Urine:Qn:
loinc:48168-9 ! 5-Hydroxyindoleacetate:PrThr:Pt:Urine:Ord:
loinc:11145-0 ! 5-Hydroxyindoleacetate/Creatinine:MRto:Pt:Urine:Qn:
loinc:1694-9 ! 5-Hydroxyindoleacetate:MCnc:Pt:Urine:Qn:
loinc:47545-9 ! 5-Hydroxyindoleacetate/Creatinine:SRto:24H:Urine:Qn:
loinc:14573-0 ! 5-Hydroxyindoleacetate:SRat:24H:Urine:Qn:

Semantic Similarity (Term Wise)

[16]:

comploinc similarity --help

Usage: runoak similarity [OPTIONS] [TERMS]...

  All by all similarity

  This calculates a similarity matrix for two sets of terms.

  Input sets of a terms can be specified in different ways:

  - via a file - via explicit lists of terms or queries

  Example:

      runoak -i hp.db all-similarity -p i --set1-file HPO-TERMS1 --set2-file
      HPO-TERMS2 -O csv

  This will compare every term in TERMS1 vs TERMS2

  Alternatively standard OAK term queries can be used, with "@" separating the
  two lists

  Example:

      runoak -i hp.db all-similarity -p i TERM_1 TERM_2 ... TERM_N @ TERM_N+1
      ... TERM_M

  The .all term syntax can be used to select all terms in an ontology

  Example:

      runoak -i ma.db all-similarity -p i,p .all @ .all

  This can be mixed with other term selectors; for example to calculate the
  similarity of "neuron" vs all terms in CL:

      runoak -i cl.db all-similarity -p i,p .all @ neuron

  An example pipeline to do all by all over all phenotypes in HPO:

  Explicit:

      runoak -i hp.db descendants -p i HP:0000118 > HPO     runoak -i hp.db
      all-similarity -p i --set1-file HPO --set2-file HPO -O csv -o
      RESULTS.tsv

  The same thing can be done more compactly with term queries:

      runoak -i hp.db all-similarity -p i .desc//p=i HP:0000118 @ .desc//p=i
      HP:0000118

Options:
  -p, --predicates TEXT         A comma-separated list of predicates
  --set1-file TEXT              ID file for set1
  --set2-file TEXT              ID file for set2
  --jaccard-minimum FLOAT       Minimum value for jaccard score
  --ic-minimum FLOAT            Minimum value for information content
  -o, --output TEXT             path to output
  --main-score-field TEXT       Score used for summarization  [default:
                                phenodigm_score]
  --autolabel / --no-autolabel  If set, results will automatically have labels
                                assigned  [default: autolabel]
  -O, --output-type TEXT        Desired output type
  --help                        Show this message and exit.

[15]:

comploinc similarity loinc:2341-6 @ loinc:3134-4

ancestor_id: loinc:LP65098-3
ancestor_information_content: 6.066089190457772
ancestor_label: Sugar
jaccard_similarity: 0.5
object_id: loinc:3134-4
object_label: 'Xylose:MCnc:Pt:Bld:Qn:'
phenodigm_score: 1.7415638361050352
subject_id: loinc:2341-6
subject_label: Glucose:MCnc:Pt:Bld:Qn:Test strip manual

Value Sets

The COMPLoinc project doesn’t define any value sets. Here we just use two random hardcoded ones for illustration purposes

[2]:

comploinc info .idfile input/valueset1.txt

loinc:5914-7 ! Glucose:PrThr:Pt:Bld:Ord:Test strip
loinc:2339-0 ! Glucose:MCnc:Pt:Bld:Qn:
loinc:50216-1 ! Glucose^6th specimen:MCnc:Pt:Ser/Plas:Qn:
loinc:6777-7 ! Glucose:MCnc:Pt:Ser/Plas:Qn:
loinc:77145-1 ! Glucose^post CFst:SCnc:Pt:Ser/Plas/Bld:Qn:
loinc:54085-6 ! Galactose:SCnc:Pt:Bld.dot:Qn:
loinc:50218-7 ! Glucose^9th specimen:MCnc:Pt:Ser/Plas:Qn:
loinc:25426-8 ! Galactose:SCnc:Pt:Ser/Plas:Qn:
loinc:32016-8 ! Glucose:MCnc:Pt:BldC:Qn:
loinc:77135-2 ! Glucose:SCnc:Pt:Ser/Plas/Bld:Qn:
loinc:2307-7 ! Galactose:MCnc:Pt:Bld:Qn:

[3]:

comploinc info .idfile input/valueset2.txt

loinc:54495-7 ! Glucose^post dialysis:SCnc:Pt:Ser/Plas:Qn:
loinc:2308-5 ! Galactose:MCnc:Pt:Ser/Plas:Qn:
loinc:76629-5 ! Glucose^post CFst:SCnc:Pt:Bld:Qn:
loinc:27353-2 ! Estimated average glucose:MCnc:Pt:Bld:Qn:Estimated from glycated hemoglobin
loinc:2552-8 ! Lactose:MCnc:Pt:Ser/Plas:Qn:
loinc:12611-0 ! Glucose^4H specimen:MCnc:Pt:Ser/Plas:Qn:
loinc:51596-5 ! Glucose:SCnc:Pt:BldC:Qn:
loinc:93791-2 ! Glucose:MCnc:Stdy^mean:Ser/Plas:Qn:
loinc:50215-3 ! Glucose^5th specimen:MCnc:Pt:Ser/Plas:Qn:
loinc:50208-8 ! Glucose^10th specimen:MCnc:Pt:Ser/Plas:Qn:
loinc:3134-4 ! Xylose:MCnc:Pt:Bld:Qn:
loinc:29999-0 ! Xylose:MCnc:Pt:Ser/Plas:Qn:
loinc:5914-7 ! Glucose:PrThr:Pt:Bld:Ord:Test strip
loinc:2339-0 ! Glucose:MCnc:Pt:Bld:Qn:
loinc:50216-1 ! Glucose^6th specimen:MCnc:Pt:Ser/Plas:Qn:
loinc:6777-7 ! Glucose:MCnc:Pt:Ser/Plas:Qn:
loinc:77145-1 ! Glucose^post CFst:SCnc:Pt:Ser/Plas/Bld:Qn:

[2]:

comploinc termset-similarity .idfile input/valueset1.txt @ .idfile input/valueset2.txt -o output/sim-out.yaml

[4]:

!head -20 output/sim-out.yaml

average_score: 9.623542661061256
best_score: 13.738514532429267
object_best_matches:
  loinc:12611-0:
    match_source: loinc:12611-0
    match_source_label: 'Glucose^4H specimen:MCnc:Pt:Ser/Plas:Qn:'
    match_target: loinc:2339-0
    match_target_label: 'Glucose:MCnc:Pt:Bld:Qn:'
    score: 6.519346011967107
    similarity:
      ancestor_id: loinc:LP14635-4
      ancestor_information_content: 6.519346011967107
      ancestor_label: Glucose
      jaccard_similarity: 0.84
      object_id: loinc:12611-0
      phenodigm_score: 2.3401390236591437
      subject_id: loinc:2339-0
  loinc:2308-5:
    match_source: loinc:2308-5
    match_source_label: 'Galactose:MCnc:Pt:Ser/Plas:Qn:'

Logical Definitions

Currently these are invisible - best way to fix this is to address:

https://github.com/loinc/comp-loinc/issues/17

[9]:

comploinc logical-definitions loinc:14573-0 loinc:47545-9

[10]:

comploinc lexmatch --help

Usage: runoak lexmatch [OPTIONS] [TERMS]...

  Performs lexical matching between pairs of terms in one more more
  ontologies.

  Examples:

      runoak -i foo.obo lexmatch -o foo.sssom.tsv

  In this example, the input ontology file is assumed to contain all pairs of
  terms to be mapped.

  It is more common to map between all pairs of terms in two ontology files.
  In this case, you can merge the ontologies using a tool like ROBOT; or,  to
  avoid a merge preprocessing step, use the --addl (-a) option to specify a
  second ontology file.

      runoak -i foo.obo --add bar.obo lexmatch -o foo.sssom.tsv

  By default, this command will compare all terms in all ontologies. You can
  use the OAK term query syntax to pass in the set of all terms to be
  compared.

  For example, to compare all terms in union of FOO and BAR namespaces:

      runoak -i foo.obo --add bar.obo lexmatch -o foo.sssom.tsv i^FOO: i^BAR:

  All members of the set are compared (including FOO to FOO matches and BAR to
  BAR matches), omitting trivial reciprocal matches.

  Use an "@" separator between two queries to feed in two explicit sets:

      runoak -i foo.obo --add bar.obo lexmatch -o foo.sssom.tsv i^FOO: @
      i^BAR:

  ALGORITHM: lexmatch implements a simple algorithm:

  - create a lexical index, keyed by normalized strings of labels, synonyms -
  report all pairs of entities that have the same key

  The lexical index can be exported (in native YAML) using -L:

      runoak -i foo.obo lexmatch -L foo.index.yaml -o foo.sssom.tsv

  Note: if you run the above command a second time it will be faster as the
  index will be reused.

  RULES: Using custom rules:

      runoak  -i foo.obo lexmatch -R match_rules.yaml -L foo.index.yaml -o
      foo.sssom.tsv

  Full documentation:

  - https://incatools.github.io/ontology-access-
  kit/src/oaklib.utilities.lexical.lexical_indexer.html# module-
  oaklib.utilities.lexical.lexical_indexer

Options:
  -R, --rules-file TEXT           path to rules file. Conforms to
                                  rules_datamodel.        e.g.
                                  https://github.com/INCATools/ontology-
                                  access-
                                  kit/blob/main/tests/input/matcher_rules.yaml
  --add-labels / --no-add-labels  Populate empty labels with URI fragments or
                                  CURIE local IDs, for ontologies that use
                                  semantic IDs  [default: no-add-labels]
  -L, --lexical-index-file TEXT   path to lexical index. This is recreated
                                  each time unless --no-recreate is passed
  --recreate / --no-recreate      if true and lexical index is specified,
                                  always recreate, otherwise load from index
                                  [default: recreate]
  -o, --output FILENAME           Output file, e.g. obo file
  --help                          Show this message and exit.

Lexical Matching

[11]:

comploinc -a sqlite:obo:uberon lexmatch -L output/loinc-uberon-lexical-index.yaml -o output/loinc-uberon.sssom.tsv i^UBERON: @ i^loinc:

WARNING:root:Skipping <urn:swrl#A> as it is not a valid CURIE
WARNING:root:Skipping <urn:swrl#B> as it is not a valid CURIE
WARNING:root:Skipping <urn:swrl#C> as it is not a valid CURIE
WARNING:root:Skipping <urn:swrl#D> as it is not a valid CURIE
WARNING:root:Skipping <urn:swrl#a1> as it is not a valid CURIE
WARNING:root:Skipping <urn:swrl#a2> as it is not a valid CURIE
WARNING:root:Skipping <urn:swrl#d> as it is not a valid CURIE
WARNING:root:Skipping <urn:swrl#e> as it is not a valid CURIE
WARNING:root:Skipping <urn:swrl#eff> as it is not a valid CURIE
WARNING:root:Skipping <urn:swrl#g1> as it is not a valid CURIE
WARNING:root:Skipping <urn:swrl#g2> as it is not a valid CURIE
WARNING:root:Skipping <urn:swrl#in> as it is not a valid CURIE
WARNING:root:Skipping <urn:swrl#mf2> as it is not a valid CURIE
WARNING:root:Skipping <urn:swrl#mf> as it is not a valid CURIE
WARNING:root:Skipping <urn:swrl#p> as it is not a valid CURIE
WARNING:root:Skipping <urn:swrl#w> as it is not a valid CURIE
WARNING:root:Skipping <urn:swrl#x> as it is not a valid CURIE
WARNING:root:Skipping <urn:swrl#y> as it is not a valid CURIE
WARNING:root:Skipping <urn:swrl#z> as it is not a valid CURIE
WARNING:root:Skipping <urn:swrl:var#x> as it is not a valid CURIE
WARNING:root:Skipping <urn:swrl:var#y> as it is not a valid CURIE
WARNING:root:Skipping <urn:swrl:var#z> as it is not a valid CURIE

[12]:

import pandas as pd

[13]:

df = pd.read_csv("output/loinc-uberon.sssom.tsv", sep="\t", comment="#")
df

[13]:

	subject_id	subject_label	predicate_id	object_id	object_label	mapping_justification	mapping_tool	subject_match_field	object_match_field	match_string
0	UBERON:0000004	nose	skos:closeMatch	loinc:LP7443-7	Nose	semapv:LexicalMatching	oaklib	oio:hasExactSynonym	rdfs:label	nose
1	UBERON:0000004	nose	skos:closeMatch	loinc:LP7443-7	Nose	semapv:LexicalMatching	oaklib	rdfs:label	rdfs:label	nose
2	UBERON:0000014	zone of skin	skos:closeMatch	loinc:LP36760-4	Skin	semapv:LexicalMatching	oaklib	oio:hasExactSynonym	rdfs:label	skin
3	UBERON:0000019	camera-type eye	skos:closeMatch	loinc:LP7797-6	EYE	semapv:LexicalMatching	oaklib	oio:hasBroadSynonym	rdfs:label	eye
4	UBERON:0000019	camera-type eye	skos:closeMatch	loinc:LP7218-3	Eye	semapv:LexicalMatching	oaklib	oio:hasBroadSynonym	rdfs:label	eye
...	...	...	...	...	...	...	...	...	...	...
74	UBERON:2000673	hypobranchial artery	skos:closeMatch	loinc:LP28800-8	HA	semapv:LexicalMatching	oaklib	oio:hasExactSynonym	rdfs:label	ha
75	UBERON:3011048	genital system	skos:closeMatch	loinc:LP7555-8	Reproductive system	semapv:LexicalMatching	oaklib	oio:hasBroadSynonym	rdfs:label	reproductive system
76	UBERON:3011048	genital system	skos:closeMatch	loinc:LP7264-7	Genitalia	semapv:LexicalMatching	oaklib	oio:hasBroadSynonym	rdfs:label	genitalia
77	UBERON:6110636	insect adult cerebral ganglion	skos:closeMatch	loinc:LP7084-9	Brain	semapv:LexicalMatching	oaklib	oio:hasRelatedSynonym	rdfs:label	brain
78	UBERON:8420000	hair of scalp	skos:closeMatch	loinc:LP7280-3	Hair	semapv:LexicalMatching	oaklib	oio:hasBroadSynonym	rdfs:label	hair

79 rows × 10 columns

[ ]: