OAK statistics command
This notebook is intended as a supplement to the main OAK CLI docs.
This notebook provides examples for the statistics command, which can be used to calculate basic descriptive statistics for an ontology
Help Option
You can get help on any OAK command using --help
[1]:
!runoak statistics --help
Usage: runoak statistics [OPTIONS] [BRANCHES]...
Shows all descriptive/summary statistics
Example: ------- runoak -i sqlite:obo:pr statistics
By default, this will show combined summary statistics for all terms
You can also break down the statistics in two ways:
- by a collection of branch roots
- by a metadata property (e.g. oio:hasOBONamespace, rdfs:isDefinedBy)
- by prefix (e.g. GO, PR, CL, OBI)
Example: ------- runoak -i sqlite:obo:pr statistics -p
oio:hasOBONamespace
Note: the oio:hasOBONamespace is *not* the same as the ID prefix, it is a
field that is used by a subset of ontologies to partition classes into broad
groupings, similar to subsets. Its use is non-standard, yet a lot of
ontologies use this as the main partitioning mechanism.
A note on bundled ontologies:
The standard release many OBO ontologies "bundles" parts of other ontologies
(formally, the release product includes a merged imports closure of import
modules). This can complicate generation of statistics. A naive count of all
classes in the main OBI release will include not only "native" OBI classes,
but also classes from other ontologies that are bundled in the release.
For bundled ontologies, we recommend some kind of partitioning, such as via
defined roots, or via the CURIE prefix, using the ``--group-by-prefix``
option.
Output formats:
The recommended output types for this command are yaml, json, or csv. The
default output type is yaml, following the SummaryStatistics data model.
This is naturally nested, as the statistics includes faceted groupings (e.g.
edge counts are broken down by predicate). When specifying a flat format
like csv, this is flattened into a single table, with dynamic column names.
Change statistics:
You can optionally combine the ontology statistics with a change summary
relative to another ontology, using the ``--compare-with`` option.
Example: ------- runoak -i v2.obo statistics --group-by-obo-namespace
--compare-with v1.obo
This will also include change stats broken down by KGCL change types. If a
group-by option is specified, these will be grouped accordingly.
Python API:
https://incatools.github.io/ontology-access-kit/interfaces/summary-
statistics
Data model:
https://w3id.org/oak/summary-statistics
Options:
-O, --output-type [obo|obojson|ofn|rdf|json|yaml|fhirjson|csv|tsv|nl]
Desired output type
--group-by-property TEXT group summaries by a metadata property, e.g.
rdfs:isDefinedBy
--group-by-obo-namespace / --no-group-by-obo-namespace
shortcut for --group-by-property
oio:hasOBONamespace (note this is distinct
from the ID namespace) [default: no-group-
by-obo-namespace]
--group-by-prefix / --no-group-by-prefix
shortcut for --group-by-property sh:prefix.
Groups by the prefix of the CURIE [default:
no-group-by-prefix]
--group-by-defined-by / --no-group-by-defined-by
shortcut for --group-by-property
rdfs:isDefinedBy. This may be inferred from
prefix if not set explicitly [default: no-
group-by-defined-by]
--include-residuals / --no-include-residuals
If true include an OTHER category for terms
that do not have the property
-X, --compare-with TEXT Compare with another ontology
-P, --has-prefix TEXT filter based on a prefix, e.g. OBI
-o, --output FILENAME Output file, e.g. obo file
--help Show this message and exit.
Set up an alias
For convenience we will set up some aliases for use in this notebook
[18]:
alias chebi runoak -i sqlite:obo:chebi
Calculating summary statistics (default YAML output)
We can calculate the summary stats using the statistics command. The output is quite lengthy, so we will use --output (-o) to direct to a yamml file:
[19]:
chebi statistics -o output/chebi.stats.yaml
WARNING:root:bad mapping: KEGG_COMPOUND
WARNING:root:bad mapping: IUPAC
WARNING:root:bad mapping: ChemIDplus
WARNING:root:bad mapping: UniProt
WARNING:root:bad mapping: DrugCentral
WARNING:root:bad mapping: LINCS
WARNING:root:bad mapping: KEGG_DRUG
WARNING:root:bad mapping: ChEBI
WARNING:root:bad mapping: ChEMBL
WARNING:root:bad mapping: DrugBank
WARNING:root:bad mapping: WHO_MedNet
WARNING:root:bad mapping: PDBeChem
WARNING:root:bad mapping: NIST_Chemistry_WebBook
WARNING:root:bad mapping: PPDB
WARNING:root:bad mapping: LIPID_MAPS
WARNING:root:bad mapping: IUPHAR
WARNING:root:bad mapping: HMDB
WARNING:root:bad mapping: SUBMITTER
WARNING:root:bad mapping: MetaCyc
WARNING:root:bad mapping: JCBN
WARNING:root:bad mapping: GlyTouCan
WARNING:root:bad mapping: KNApSAcK
WARNING:root:bad mapping: IUBMB
WARNING:root:bad mapping: CBN
WARNING:root:bad mapping: Alan_Wood's_Pesticides
WARNING:root:bad mapping: GlyGen
WARNING:root:bad mapping: KEGG_GLYCAN
WARNING:root:bad mapping: RESID
WARNING:root:bad mapping: PubChem
WARNING:root:bad mapping: FooDB
WARNING:root:bad mapping: VSDB
WARNING:root:bad mapping: UM-BBD
WARNING:root:bad mapping: MolBase
WARNING:root:bad mapping: COMe
WARNING:root:bad mapping: Beilstein
WARNING:root:bad mapping: Patent
WARNING:root:bad mapping: PDB
WARNING:root:bad mapping: SMID
Note CHEBI has a lot of bad xrefs, hence the output
Exploring the output
Let’s look at the top of the YAML file:
[20]:
!head -50 output/chebi.stats.yaml
id: AllOntologies
ontologies:
- id: obo:chebi.owl
version: obo:chebi/231/chebi.owl
was_generated_by:
started_at_time: '2024-03-26T17:29:56.627143'
was_associated_with: OAK
acted_on_behalf_of: cjm
class_count: 217549
deprecated_class_count: 18650
non_deprecated_class_count: 198899
class_count_with_text_definitions: 53575
class_count_without_text_definitions: 163974
object_property_count: 10
annotation_property_count: 37
named_individual_count: 0
subset_count: 3
rdf_triple_count: 6860047
subclass_of_axiom_count: 368285
equivalent_classes_axiom_count: 0
edge_count_by_predicate:
BFO:0000051:
facet: BFO:0000051
filtered_count: 4029
RO:0000087:
facet: RO:0000087
filtered_count: 43636
obo:chebi#has_functional_parent:
facet: obo:chebi#has_functional_parent
filtered_count: 19632
obo:chebi#has_parent_hydride:
facet: obo:chebi#has_parent_hydride
filtered_count: 1799
obo:chebi#is_conjugate_acid_of:
facet: obo:chebi#is_conjugate_acid_of
filtered_count: 8484
obo:chebi#is_conjugate_base_of:
facet: obo:chebi#is_conjugate_base_of
filtered_count: 8484
obo:chebi#is_enantiomer_of:
facet: obo:chebi#is_enantiomer_of
filtered_count: 2754
obo:chebi#is_substituent_group_from:
facet: obo:chebi#is_substituent_group_from
filtered_count: 1287
obo:chebi#is_tautomer_of:
facet: obo:chebi#is_tautomer_of
filtered_count: 1886
rdfs:subClassOf:
facet: rdfs:subClassOf
Like all objects produced by OAK, there is a data dictionary / data model. The ontology stats one is https://w3id.org/oak/summary-statistics, you can use this link to browse documentation etc.
A well defined data dictionary is necessary for communicating aggregate statistics accurately. Often when ontologies are reported informally, it’s ambiguous whether number of terms means:
number of classes, classes plus relationship types, or classes plus some other elements
including or excluding deprecated (obsolete) entities
The OAK summary statistics data dictionary aims to provide a standard for ontology reporting.
YAML allows for nesting which is a natural way to group things; for example:
edge_count_by_predicate:
BFO:0000051:
facet: BFO:0000051
filtered_count: 4003
RO:0000087:
facet: RO:0000087
filtered_count: 43082
This says that there are 4003 part-of (BFO:0000050) and 43082 has-role (RO:00000087) relationships.
See the OAK guide to relationships to understand more.
Mapping Stats
Further on in the YAML we can see mapping stats. See (https://w3id.org/ssssom)[https://w3id.org/ssssom] to understand the OAK mapping data model.
These are broken down
by mapping predicate (for many ontologies this is only
oio:hasDbXref)my mapping object source (i.e. the database or ontology that is mapped to)
[21]:
!grep -A40 ^mapping_statement_count output/chebi.stats.yaml
mapping_statement_count_by_predicate:
oio:hasDbXref:
facet: oio:hasDbXref
filtered_count: 345271
mapping_statement_count_by_object_source:
BFO:
facet: BFO
filtered_count: 1
RO:
facet: RO
filtered_count: 1
KNApSAcK:
facet: KNApSAcK
filtered_count: 5185
KEGG:
facet: KEGG
filtered_count: 22228
CAS:
facet: CAS
filtered_count: 28938
KEGG_COMPOUND:
facet: KEGG_COMPOUND
filtered_count: 19870
Beilstein:
facet: Beilstein
filtered_count: 9187
IUPAC:
facet: IUPAC
filtered_count: 61013
ChemIDplus:
facet: ChemIDplus
filtered_count: 33383
UniProt:
facet: UniProt
filtered_count: 16047
LINCS:
facet: LINCS
filtered_count: 41392
Drug_Central:
facet: Drug_Central
filtered_count: 3784
DrugCentral:
facet: DrugCentral
filtered_count: 6202
Wikipedia:
--
mapping_statement_count_subject_by_object_source:
BFO:
facet: BFO
filtered_count: 1
RO:
facet: RO
filtered_count: 1
KNApSAcK:
facet: KNApSAcK
filtered_count: 5091
KEGG:
facet: KEGG
filtered_count: 20233
CAS:
facet: CAS
filtered_count: 28615
KEGG_COMPOUND:
facet: KEGG_COMPOUND
filtered_count: 19870
Beilstein:
facet: Beilstein
filtered_count: 8704
IUPAC:
facet: IUPAC
filtered_count: 61013
ChemIDplus:
facet: ChemIDplus
filtered_count: 33383
UniProt:
facet: UniProt
filtered_count: 16047
LINCS:
facet: LINCS
filtered_count: 41389
Drug_Central:
facet: Drug_Central
filtered_count: 3783
DrugCentral:
facet: DrugCentral
filtered_count: 6202
Wikipedia:
As expected, CHEBI does not make use of SKOS mapping predicates, and mappings are dominated by databases like KEGG, CAS.
TSV Output
YAML is not a very natural format for doing further data science or statistical processing.
FOr that we can use the csv option (which actually defaults to tsv…)
[9]:
chebi statistics -o output/chebi.stats.tsv -O csv
WARNING:root:bad mapping: KEGG_COMPOUND
WARNING:root:bad mapping: IUPAC
WARNING:root:bad mapping: ChemIDplus
WARNING:root:bad mapping: UniProt
WARNING:root:bad mapping: DrugCentral
WARNING:root:bad mapping: LINCS
WARNING:root:bad mapping: KEGG_DRUG
WARNING:root:bad mapping: ChEBI
WARNING:root:bad mapping: ChEMBL
WARNING:root:bad mapping: DrugBank
WARNING:root:bad mapping: WHO_MedNet
WARNING:root:bad mapping: PDBeChem
WARNING:root:bad mapping: NIST_Chemistry_WebBook
WARNING:root:bad mapping: LIPID_MAPS
WARNING:root:bad mapping: IUPHAR
WARNING:root:bad mapping: HMDB
WARNING:root:bad mapping: SUBMITTER
WARNING:root:bad mapping: MetaCyc
WARNING:root:bad mapping: JCBN
WARNING:root:bad mapping: GlyTouCan
WARNING:root:bad mapping: KNApSAcK
WARNING:root:bad mapping: IUBMB
WARNING:root:bad mapping: EMBL
WARNING:root:bad mapping: CBN
WARNING:root:bad mapping: Alan_Wood's_Pesticides
WARNING:root:bad mapping: GlyGen
WARNING:root:bad mapping: PPDB
WARNING:root:bad mapping: KEGG_GLYCAN
WARNING:root:bad mapping: RESID
WARNING:root:bad mapping: PubChem
WARNING:root:bad mapping: FooDB
WARNING:root:bad mapping: VSDB
WARNING:root:bad mapping: UM-BBD
WARNING:root:bad mapping: MolBase
WARNING:root:bad mapping: COMe
WARNING:root:bad mapping: EBI_Industry_Programme
WARNING:root:bad mapping: EuroFIR
WARNING:root:bad mapping: Beilstein
WARNING:root:bad mapping: Patent
WARNING:root:bad mapping: PDB
WARNING:root:bad mapping: SMID
To illustrate this we will use pandas:
[11]:
import pandas as pd
df = pd.read_csv("output/chebi.stats.tsv", sep="\t")
df
[11]:
| id | compared_with | agents | class_count | deprecated_class_count | non_deprecated_class_count | class_count_with_text_definitions | class_count_without_text_definitions | object_property_count | annotation_property_count | ... | mapping_statement_count_subject_by_object_source_CTX | mapping_statement_count_subject_by_object_source_SMID | class_count_by_subset_1_STAR | class_count_by_subset_2_STAR | class_count_by_subset_3_STAR | was_generated_by_started_at_time | was_generated_by_was_associated_with | was_generated_by_acted_on_behalf_of | ontologies_id | ontologies_version | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | AllOntologies | NaN | NaN | 185295 | 18628 | 166667 | 53049 | 132246 | 10 | 37 | ... | 3 | 307 | 2945 | 102919 | 60803 | 2024-03-26T17:07:33.778117 | OAK | cjm | obo:chebi.owl | obo:chebi/226/chebi.owl |
1 rows × 177 columns
This format is useful if you have multiple ontologies (see later). But for a single ontology it’s more convenient to melt this:
[13]:
mdf = df.melt(var_name='Property', value_name='Value')
mdf[0:40]
[13]:
| Property | Value | |
|---|---|---|
| 0 | id | AllOntologies |
| 1 | compared_with | NaN |
| 2 | agents | NaN |
| 3 | class_count | 185295 |
| 4 | deprecated_class_count | 18628 |
| 5 | non_deprecated_class_count | 166667 |
| 6 | class_count_with_text_definitions | 53049 |
| 7 | class_count_without_text_definitions | 132246 |
| 8 | object_property_count | 10 |
| 9 | annotation_property_count | 37 |
| 10 | named_individual_count | 0 |
| 11 | subset_count | 3 |
| 12 | rdf_triple_count | 6158555 |
| 13 | subclass_of_axiom_count | 330989 |
| 14 | equivalent_classes_axiom_count | 0 |
| 15 | entailed_edge_count_by_predicate | {} |
| 16 | distinct_synonym_count | 332744 |
| 17 | synonym_statement_count | 346486 |
| 18 | class_count_by_category | {} |
| 19 | contributor_summary | {} |
| 20 | change_summary | {} |
| 21 | merged_class_query | 18559 |
| 22 | deprecated_property_count | 0 |
| 23 | edge_count_by_predicate_BFO:0000051 | 4003 |
| 24 | edge_count_by_predicate_RO:0000087 | 43082 |
| 25 | edge_count_by_predicate_has_functional_parent | 18664 |
| 26 | edge_count_by_predicate_has_parent_hydride | 1764 |
| 27 | edge_count_by_predicate_is_conjugate_acid_of | 8434 |
| 28 | edge_count_by_predicate_is_conjugate_base_of | 8434 |
| 29 | edge_count_by_predicate_is_enantiomer_of | 2728 |
| 30 | edge_count_by_predicate_is_substituent_group_from | 1284 |
| 31 | edge_count_by_predicate_is_tautomer_of | 1884 |
| 32 | edge_count_by_predicate_rdfs:subClassOf | 240712 |
| 33 | edge_count_by_predicate_rdfs:subPropertyOf | 6 |
| 34 | synonym_statement_count_by_predicate_hasExactS... | 100585 |
| 35 | synonym_statement_count_by_predicate_hasRelate... | 234002 |
| 36 | mapping_statement_count_by_predicate_hasDbXref | 317151 |
| 37 | mapping_statement_count_by_object_source_BFO | 1 |
| 38 | mapping_statement_count_by_object_source_RO | 1 |
| 39 | mapping_statement_count_by_object_source_KNApSAcK | 5152 |
Note this uses a very generic way of flattening the yaml so some columns make less sense out of context - e.g. the “agent” field belongs to a parent object that describes what “agent” generated the stats (TODO: this should say “oaklib”)
Multi-ontology merges
Many OBO ontologies bundle portions of other ontologies with their main release. This can be confusing! For more details see OWL Format Variants in the obook.
As an example, consider naively calculating stats for the standard release of the Cell Ontology (CL):
[14]:
!runoak -i sqlite:obo:cl statistics | head -20
WARNING:root:bad mapping: GSE137537
WARNING:root:bad mapping: 10.1007/s004180050142
WARNING:root:bad mapping: NIFSTD
WARNING:root:bad mapping: Noradrenergic_cell_group_A6&oldid=981960774
WARNING:root:bad mapping: _Chapter_3
WARNING:root:bad mapping: A12.2.15.042
WARNING:root:bad mapping: PHENOSCAPE
id: AllOntologies
ontologies:
- id: obo:cl.owl
version: obo:cl/releases/2023-09-21/cl.owl
version_info: '2023-09-21'
was_generated_by:
started_at_time: '2024-03-26T17:16:02.669245'
was_associated_with: OAK
acted_on_behalf_of: cjm
class_count: 28330
deprecated_class_count: 261
non_deprecated_class_count: 28069
class_count_with_text_definitions: 15110
class_count_without_text_definitions: 13220
object_property_count: 297
annotation_property_count: 241
named_individual_count: 18
subset_count: 63
rdf_triple_count: 681623
subclass_of_axiom_count: 44142
Looking at this you might think CL has 28k classes. In fact, this is the total number of classes in the ontology as defined by OWL, where here “ontology” means the merged product that includes bits of GO, Uberon, etc. Confusing, huh?
Ideally the OBO Foundry would move towards making base files the default, but in the absence of this, we have a few options:
Filtering by prefix (using
-P)Grouping using some property such as the prefix.
We’ll try the latter
[22]:
!runoak -i sqlite:obo:cl statistics --group-by-prefix -o output/cl.stats.grouped.tsv -O csv
[23]:
df = pd.read_csv("output/cl.stats.grouped.tsv", sep="\t")
df
[23]:
| id | compared_with | agents | class_count | deprecated_class_count | non_deprecated_class_count | class_count_with_text_definitions | class_count_without_text_definitions | object_property_count | annotation_property_count | ... | class_count_by_subset_non_informative | class_count_by_subset_organ_slim | class_count_by_subset_pheno_slim | class_count_by_subset_phenotype_rcn | class_count_by_subset_uberon_slim | class_count_by_subset_unverified_taxonomic_grouping | class_count_by_subset_upper_level | class_count_by_subset_vertebrate_core | mapping_statement_count_by_object_source_GOREL | mapping_statement_count_subject_by_object_source_GOREL | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | <http | NaN | NaN | 0 | 0 | 0 | 0 | 0 | 0 | 1 | ... | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN |
| 1 | <https | NaN | NaN | 0 | 0 | 0 | 0 | 0 | 0 | 1 | ... | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN |
| 2 | BFO | NaN | NaN | 15 | 0 | 15 | 9 | 6 | 6 | 0 | ... | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN |
| 3 | BSPO | NaN | NaN | 0 | 0 | 0 | 0 | 0 | 24 | 0 | ... | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN |
| 4 | CARO | NaN | NaN | 20 | 0 | 20 | 20 | 0 | 0 | 0 | ... | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN |
| 5 | CHEBI | NaN | NaN | 123 | 0 | 123 | 18 | 105 | 0 | 0 | ... | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN |
| 6 | CL | NaN | NaN | 2969 | 249 | 2720 | 2555 | 414 | 3 | 0 | ... | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN |
| 7 | GO | NaN | NaN | 7265 | 2 | 7263 | 7264 | 1 | 0 | 0 | ... | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN |
| 8 | IAO | NaN | NaN | 6 | 0 | 6 | 4 | 2 | 0 | 23 | ... | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN |
| 9 | NCBITaxon | NaN | NaN | 138 | 0 | 138 | 0 | 138 | 0 | 0 | ... | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN |
| 10 | OMO | NaN | NaN | 0 | 0 | 0 | 0 | 0 | 0 | 2 | ... | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN |
| 11 | PATO | NaN | NaN | 185 | 0 | 185 | 184 | 1 | 0 | 0 | ... | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN |
| 12 | PR | NaN | NaN | 748 | 0 | 748 | 747 | 1 | 0 | 0 | ... | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN |
| 13 | RO | NaN | NaN | 1 | 0 | 1 | 1 | 0 | 240 | 16 | ... | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN |
| 14 | UBERON | NaN | NaN | 4670 | 0 | 4670 | 4308 | 362 | 0 | 0 | ... | 47.0 | 136.0 | 1373.0 | 3.0 | 809.0 | 1.0 | 49.0 | 448.0 | NaN | NaN |
| 15 | cito | NaN | NaN | 0 | 0 | 0 | 0 | 0 | 0 | 1 | ... | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN |
| 16 | dce | NaN | NaN | 0 | 0 | 0 | 0 | 0 | 0 | 6 | ... | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN |
| 17 | dcterms | NaN | NaN | 0 | 0 | 0 | 0 | 0 | 0 | 7 | ... | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN |
| 18 | foaf | NaN | NaN | 0 | 0 | 0 | 0 | 0 | 0 | 2 | ... | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN |
| 19 | obo | NaN | NaN | 10 | 10 | 0 | 0 | 10 | 24 | 116 | ... | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | 2.0 | 2.0 |
| 20 | oio | NaN | NaN | 0 | 0 | 0 | 0 | 0 | 0 | 61 | ... | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN |
| 21 | owl | NaN | NaN | 0 | 0 | 0 | 0 | 0 | 0 | 1 | ... | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN |
| 22 | rdfs | NaN | NaN | 0 | 0 | 0 | 0 | 0 | 0 | 3 | ... | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN |
| 23 | skos | NaN | NaN | 0 | 0 | 0 | 0 | 0 | 0 | 1 | ... | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN |
| 24 | xsd | NaN | NaN | 0 | 0 | 0 | 0 | 0 | 0 | 0 | ... | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN |
25 rows × 512 columns
Here we can see the numbers broken down by ontology. The number of classes in the CL row is now accurate. Note of course that the other numbers don’t reflect totals for the external ontology as a whole – it’s just the number that has been merged into CL
Diff stats
You can also use --compare-with to compare stats with a different release of an ontology. Note this is effictively the same as running diff with --statistics. See diff docs for details.
[ ]: