In meerapatelmd/mOMOP: What the Package Does (One Line, Title Case)

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  message = FALSE
)

library(tidyverse)
library(mOMOP)

The mappings between the mCode Value Sets and OMOP Concepts is converted into an RDF to view and edit in Protege.

There are 3 vocabulary-level datasets in this package that require additional parsing for appropriate categorization into owl: SNOMED, ICD10CM, and LOINC. This categarization is done by assessing the source mCode value sets each belongs to.

In mCode, LOINC is used only for tumor marker testing with no overlap with any other dataset.

unique(LOINC$value_set_name)

All the valuesets represented by ICD10CM overlaps with SNOMED.

unique(ICD10CM$value_set_name)
unique(SNOMED$value_set_name)

To manage the overlaps, ICD10CM and SNOMED are combined and split by value set.

ICD10CM_SNOMED <-
  bind_rows(ICD10CM,
            SNOMED) %>%
  rubix::split_by(col = value_set_name)
names(ICD10CM_SNOMED)

The names are adjusted to following the uppercase snake format for this project.

nms <- names(ICD10CM_SNOMED)
nms <- str_remove_all(nms, "VS$")
nms <- str_replace_all(nms, "([a-z]{1})([A-Z]{1})","\\1_\\2")
nms <- toupper(nms)
names(ICD10CM_SNOMED) <- nms
names(ICD10CM_SNOMED)

The other datasets are combined into a list with LOINC renamed to TUMOR_MARKER_MEASUREMENT. Each dataset in the mcode_classes list is now divided by "Class" for the OWL class hierarchy.

other_value_sets <-
list(CANCER_STAGING = CANCER_STAGING, 
          GENOMICS = GENOMICS, 
          TUMOR_MARKER_MEASUREMENT = LOINC, 
          SPECIMEN = SPECIMEN, 
          UNITS_OF_MEASUREMENT = UNITS_OF_MEASUREMENT) 

mcode_classes <-
  c(ICD10CM_SNOMED,
    other_value_sets)

names(mcode_classes)

All the concepts in each class are converted to the chariot package's strip format to represent them as single instances in OWL. For the SPECIMEN class specifically, the original mCode code and code description are carried over as instances to provide the appropriate gap coverage in the specimen representation.

mcode_classes2 <-
mcode_classes %>%
  map(mutate_all, as.character) %>%
  bind_rows(.id = "class") %>% 
  chariot::merge_strip(into = "concept") %>%
  # Specimen domain has additional columns
  unite(
    col = "mcode_concept", 
    code, code_description, 
    sep = " ", 
    remove = FALSE,
    na.rm = TRUE
    ) %>%
  mutate(concept = coalesce(concept, Specimen, mcode_concept)) %>%
  distinct() %>%
  rubix::split_by(col = "class")

The data is written to an Excel to load when calling the data function.

file <- file.path(getwd(), "data-raw/mcode_rdf.xlsx")
if (!file.exists(file)) {
  openxlsx::write.xlsx(x = mcode_classes2,
                       file = file)
}

Data Glossary

mcode_classes3 <-
  mcode_classes2 %>%
  map(select, concept)
mcode_classes3

meerapatelmd/mOMOP documentation built on June 15, 2021, 3:01 p.m.