Matching GWAS traits to GBD causes / diseases
Isobel Beasley
2025-10-07
Last updated: 2026-03-31
Checks: 7 0
Knit directory:
genomics_ancest_disease_dispar/
This reproducible R Markdown analysis was created with workflowr (version 1.7.2). The Checks tab describes the reproducibility checks that were applied when the results were created. The Past versions tab lists the development history.
Great! Since the R Markdown file has been committed to the Git repository, you know the exact version of the code that produced these results.
Great job! The global environment was empty. Objects defined in the global environment can affect the analysis in your R Markdown file in unknown ways. For reproduciblity it’s best to always run the code in an empty environment.
The command set.seed(20220216) was run prior to running
the code in the R Markdown file. Setting a seed ensures that any results
that rely on randomness, e.g. subsampling or permutations, are
reproducible.
Great job! Recording the operating system, R version, and package versions is critical for reproducibility.
Nice! There were no cached chunks for this analysis, so you can be confident that you successfully produced the results during this run.
Great job! Using relative paths to the files within your workflowr project makes it easier to run your code on other machines.
Great! You are using Git for version control. Tracking code development and connecting the code version to the results is critical for reproducibility.
The results in this page were generated with repository version f3cf2b5. See the Past versions tab to see a history of the changes made to the R Markdown and HTML files.
Note that you need to be careful to ensure that all relevant files for
the analysis have been committed to Git prior to generating the results
(you can use wflow_publish or
wflow_git_commit). workflowr only checks the R Markdown
file, but you know if there are other scripts or data files that it
depends on. Below is the status of the Git repository when the results
were generated:
Ignored files:
Ignored: .DS_Store
Ignored: .Rproj.user/
Ignored: .venv/
Ignored: BC2GM/
Ignored: BioC.dtd
Ignored: FormatConverter.jar
Ignored: FormatConverter.zip
Ignored: analysis/.DS_Store
Ignored: ancestry_dispar_env/
Ignored: code/.DS_Store
Ignored: code/full_text_conversion/.DS_Store
Ignored: data/.DS_Store
Ignored: data/RCDCFundingSummary_01042026.xlsx
Ignored: data/cdc/
Ignored: data/cohort/
Ignored: data/epmc/
Ignored: data/europe_pmc/
Ignored: data/gbd/.DS_Store
Ignored: data/gbd/IHME-GBD_2021_DATA-d8cf695e-1.csv
Ignored: data/gbd/IHME-GBD_2023_DATA-73cc01fd-1.csv
Ignored: data/gbd/gbd_2019_california_percent_deaths.csv
Ignored: data/gbd/ihme_gbd_2019_global_disease_burden_rate_all_ages.csv
Ignored: data/gbd/ihme_gbd_2019_global_paf_rate_percent_all_ages.csv
Ignored: data/gbd/ihme_gbd_2021_global_disease_burden_rate_all_ages.csv
Ignored: data/gbd/ihme_gbd_2021_global_paf_rate_percent_all_ages.csv
Ignored: data/gwas_catalog/
Ignored: data/icd/.DS_Store
Ignored: data/icd/2025AA/
Ignored: data/icd/IHME_GBD_2019_COD_CAUSE_ICD_CODE_MAP_Y2020M10D15.XLSX
Ignored: data/icd/IHME_GBD_2019_NONFATAL_CAUSE_ICD_CODE_MAP_Y2020M10D15.XLSX
Ignored: data/icd/IHME_GBD_2021_COD_CAUSE_ICD_CODE_MAP_Y2024M05D16.XLSX
Ignored: data/icd/IHME_GBD_2021_NONFATAL_CAUSE_ICD_CODE_MAP_Y2024M05D16.XLSX
Ignored: data/icd/UK_Biobank_master_file.tsv
Ignored: data/icd/cdc_valid_icd10_Sep_23_2025.xlsx
Ignored: data/icd/cdc_valid_icd9_Sep_23_2025.xlsx
Ignored: data/icd/hp_umls_mapping.csv
Ignored: data/icd/lancet_conditions_icd10.xlsx
Ignored: data/icd/manual_disease_icd10_mappings.xlsx
Ignored: data/icd/mondo_umls_mapping.csv
Ignored: data/icd/phecode_international_version_unrolled.csv
Ignored: data/icd/phecode_to_icd10_manual_mapping.xlsx
Ignored: data/icd/semiautomatic_ICD-pheno.txt
Ignored: data/icd/semiautomatic_ICD-pheno_UKB_subset.txt
Ignored: data/icd/umls-2025AA-mrconso.zip
Ignored: doccano_venv/
Ignored: figures/
Ignored: output/.DS_Store
Ignored: output/abstracts/
Ignored: output/doccano/
Ignored: output/fulltexts/
Ignored: output/gwas_cat/
Ignored: output/gwas_cohorts/
Ignored: output/icd_map/
Ignored: output/pubmedbert_entity_predictions.csv
Ignored: output/pubmedbert_entity_predictions.jsonl
Ignored: output/pubmedbert_predictions.csv
Ignored: output/pubmedbert_predictions.jsonl
Ignored: output/supplement/
Ignored: output/text_mining_predictions/
Ignored: output/trait_ontology/
Ignored: population_description_terms.txt
Ignored: pubmedbert-cohort-ner-model/
Ignored: pubmedbert-cohort-ner/
Ignored: renv/
Ignored: spacy_venv_requirements.txt
Ignored: spacyr_venv/
Untracked files:
Untracked: code/full_text_conversion/html_to_xml.R
Untracked: code/text_mining_models/tokenise_data.py
Unstaged changes:
Modified: analysis/disease_inves_by_ancest.Rmd
Modified: analysis/replication_ancestry_bias.Rmd
Modified: analysis/specific_aims_stats.Rmd
Modified: analysis/text_for_cohort_labels.Rmd
Note that any generated files, e.g. HTML, png, CSS, etc., are not included in this status report because it is ok for generated content to have uncommitted changes.
These are the previous versions of the repository in which changes were
made to the R Markdown (analysis/gwas_to_gbd.Rmd) and HTML
(docs/gwas_to_gbd.html) files. If you’ve configured a
remote Git repository (see ?wflow_git_remote), click on the
hyperlinks in the table below to view the files as they were in that
past version.
| File | Version | Author | Date | Message |
|---|---|---|---|---|
| Rmd | f3cf2b5 | IJbeasley | 2026-03-31 | Add upset plots to test paper overlap per condition |
| html | 5f3289b | IJbeasley | 2026-03-26 | Build site. |
| Rmd | f754dc4 | IJbeasley | 2026-03-26 | Adding missed papers |
| html | 41c0bbe | IJbeasley | 2026-01-12 | Build site. |
| Rmd | 94b4ca3 | IJbeasley | 2026-01-12 | Update initial trait categorisation |
| html | bf2e581 | IJbeasley | 2026-01-05 | Build site. |
| Rmd | 11f714c | IJbeasley | 2026-01-05 | Update filtering of GWAS traits |
| html | 33cdcc6 | IJbeasley | 2026-01-03 | Build site. |
| Rmd | 6160eba | IJbeasley | 2026-01-03 | Update fixing of trait mapping |
| html | 9d0fd3f | IJbeasley | 2025-10-09 | Build site. |
| Rmd | 8e94754 | IJbeasley | 2025-10-09 | Integrating gwas traits with ICD10 |
Set up
library(dplyr)
library(stringr)
library(purrr)
library(tidyr)
library(data.table)Load Lancet priority diseases
gbd_2019 <- readxl::read_xlsx(here::here("data/icd/lancet_conditions_icd10.xlsx"))
gbd_2019 =
gbd_2019 |>
select(cause = mapped_gbd_term,
icd10 = icd_10_non_fatal,
cause_hierarchy_level = gbd_level)
gbd_2019 =
gbd_2019 |>
filter(!is.na(icd10))Prepare GBD 2019 data for merging with GWAS traits
Tidy ICD codes
Make long format, separate multiple ICD code ranges into rows
# expand multiple ICD codes into rows
gbd_2019 =
gbd_2019 |>
mutate(icd10 = str_split(icd10, ",\\s*")) |>
tidyr::unnest(icd10)
# separate ICD code ranges into start and end codes
gbd_2019 =
gbd_2019 |>
tidyr::separate_wider_delim(col = "icd10",
delim = "-",
names = c("start_icd10_code", "end_icd10_code"),
too_few = "align_end"
)
# if only one ICD code (i.e. no range), set start and end to be the same
gbd_2019 =
gbd_2019 |>
mutate(start_icd10_code =
ifelse(is.na(start_icd10_code),
end_icd10_code,
start_icd10_code)
)
gbd_2019 =
gbd_2019 |>
distinct()
head(gbd_2019)# A tibble: 6 × 4
cause start_icd10_code end_icd10_code cause_hierarchy_level
<chr> <chr> <chr> <dbl>
1 Ischemic heart disease I20 I21.6 3
2 Ischemic heart disease I21.9 I25.9 3
3 Ischemic heart disease Z82.4 Z82.49 3
4 Ischemic stroke G45 G46.8 4
5 Ischemic stroke I63 I63.9 4
6 Ischemic stroke I65 I66.9 4Where icd10 code ranges are the same, but missing decimal places, add .9 to end code (to help with filtering ranges)
gbd_2019 |>
filter(start_icd10_code == end_icd10_code &
!grepl(".", start_icd10_code, fixed = TRUE)
) |>
arrange(start_icd10_code) |>
head()# A tibble: 5 × 4
cause start_icd10_code end_icd10_code cause_hierarchy_level
<chr> <chr> <chr> <dbl>
1 Diarrhoeal diseases A09 A09 3
2 Lower respiratory infec… A70 A70 3
3 Tracheal, bronchus, and… C33 C33 3
4 Maternal disorders N96 N96 3
5 HIV/AIDS Z21 Z21 3gbd_2019 =
gbd_2019 |>
mutate(end_icd10_code =
ifelse(
start_icd10_code == end_icd10_code &
!grepl(".", start_icd10_code, fixed = TRUE),
paste0(end_icd10_code, ".9"),
end_icd10_code
)
)Seperate numeric and letter parts of ICD codes (to help with filtering ranges)
# replace O9A513 with O99.9
gbd_2019 =
gbd_2019 |>
mutate(end_icd10_code =
ifelse(end_icd10_code == "O9A513",
"O99.9",
end_icd10_code)
)
# replace Z3A49 with Z39.2
gbd_2019 =
gbd_2019 |>
mutate(end_icd10_code =
ifelse(end_icd10_code == "Z3A49",
"Z39.2",
end_icd10_code)
)
# check start_icd10_code and end_icd10_code start with the same letter
# if not, need to fix these rows
gbd_2019 |>
filter(str_extract(start_icd10_code, "^[A-Z]") !=
str_extract(end_icd10_code, "^[A-Z]")
)# A tibble: 0 × 4
# ℹ 4 variables: cause <chr>, start_icd10_code <chr>, end_icd10_code <chr>,
# cause_hierarchy_level <dbl>gbd_2019 =
gbd_2019 |>
mutate(icd10_code_letter = str_extract(start_icd10_code, "^[A-Z]"))
gbd_2019 =
gbd_2019 |>
mutate(start_icd10_code_num = as.numeric(str_remove(start_icd10_code,
"^[A-Z]")
),
end_icd10_code_num = as.numeric(str_remove(end_icd10_code,
"^[A-Z]")
)
)Merge GWAS traits with GBD causes
Load Disease Mapping
disease_mapping <- data.table::fread(
here::here("output/icd_map/gwas_disease_to_icd10_mapping.csv")
)
disease_mapping =
disease_mapping |>
mutate(icd10_code = sub("^([^-]+)-\\1$", "\\1", icd10_code))
disease_mapping =
disease_mapping |>
mutate(icd10_code_letter = str_extract(icd10_code, "^[A-Z]")) |>
mutate(icd10_code_num = as.numeric(str_remove(icd10_code, "^[A-Z]")))
disease_mapping |>
filter(is.na(icd10_code_num)) |>
nrow()[1] 3382disease_mapping =
disease_mapping |>
filter(!is.na(icd10_code_num))
disease_mapping =
disease_mapping |>
mutate(icd10_code_num = as.numeric(icd10_code_num))Join disease mapping with GBD 2019 data
disease_mapping_with_cause <- disease_mapping |>
rowwise() |>
mutate(
cause =
list(
gbd_2019$cause[
which(icd10_code_num >= gbd_2019$start_icd10_code_num &
icd10_code_num <= gbd_2019$end_icd10_code_num &
icd10_code_letter == gbd_2019$icd10_code_letter)
]),
cause_hierarchy_level =
list(
gbd_2019$cause_hierarchy_level[
which(icd10_code_num >= gbd_2019$start_icd10_code_num &
icd10_code_num <= gbd_2019$end_icd10_code_num &
icd10_code_letter == gbd_2019$icd10_code_letter)
])
) |>
ungroup()
disease_mapping_with_cause =
disease_mapping_with_cause |>
tidyr::unnest(c(cause, cause_hierarchy_level),
keep_empty = TRUE
)# K76.1 ULMS mapping to chronic liver disease, but is actually about downstream consequences of
# heart failure/disease
disease_mapping_with_cause |>
filter(icd10_code == "K76.1" & grepl("UMLS", icd10_code_origin)) |>
select(`DISEASE/TRAIT`, cause) |>
distinct() |>
head()# A tibble: 6 × 2
`DISEASE/TRAIT` cause
<chr> <chr>
1 coronary heart disease Cirrhosis…
2 coronary artery disease Cirrhosis…
3 myocardial infarction in coronary artery disease Cirrhosis…
4 coronary heart disease (snp x snp interaction) Cirrhosis…
5 coronary heart disease event reduction (statin therapy interaction) Cirrhosis…
6 sudden cardiac arrest in coronary artery disease Cirrhosis…# remove this mapping
disease_mapping_with_cause =
disease_mapping_with_cause |>
filter(!(icd10_code == "K76.1" & grepl("UMLS", icd10_code_origin)))print("Number of unique studies mapped to a cause:")[1] "Number of unique studies mapped to a cause:"disease_mapping_with_cause |>
filter(!is.na(cause)) |>
pull(PUBMED_ID) |>
unique() |>
length()[1] 1001print("Number of unique studies mapped to non-communicable diseases:")[1] "Number of unique studies mapped to non-communicable diseases:"# filter out infectious diseases
disease_mapping_with_cause |>
dplyr::filter(!cause %in% c("HIV/AIDS",
"Tuberculosis",
"Malaria",
"Lower respiratory infections",
"Diarrhoeal diseases",
"Neonatal disorders",
"Tetanus",
"Diphtheria",
"Pertussis" ,
"Measles",
"Maternal disorders")) |>
filter(!is.na(cause)) |>
pull(PUBMED_ID) |>
unique() |>
length()[1] 828Check overlap (upset plot) of studies mapped to different causes
Between type 1 diabetes, type 2 diabetes, and diabetes mellitus
diabetes_studies = disease_mapping_with_cause |>
filter(cause %in% c("Type 1 diabetes mellitus",
"Type 2 diabetes mellitus",
"Diabetes mellitus",
"Chronic kidney disease due to diabetes mellitus type 2",
"Chronic kidney disease due to diabetes mellitus type 1")) |>
select(PUBMED_ID, cause) |>
distinct()
library(ComplexHeatmap)Loading required package: grid========================================
ComplexHeatmap version 2.18.0
Bioconductor page: http://bioconductor.org/packages/ComplexHeatmap/
Github page: https://github.com/jokergoo/ComplexHeatmap
Documentation: http://jokergoo.github.io/ComplexHeatmap-reference
If you use it in published research, please cite either one:
- Gu, Z. Complex Heatmap Visualization. iMeta 2022.
- Gu, Z. Complex heatmaps reveal patterns and correlations in multidimensional
genomic data. Bioinformatics 2016.
The new InteractiveComplexHeatmap package can directly export static
complex heatmaps into an interactive Shiny app with zero effort. Have a try!
This message can be suppressed by:
suppressPackageStartupMessages(library(ComplexHeatmap))
========================================# create a binary matrix indicating which studies are mapped to which causes
diabetes_studies_wide = diabetes_studies |>
tidyr::pivot_wider(names_from = cause, values_from = cause, values_fill = NA) |>
select(-PUBMED_ID) |>
mutate_all(~ ifelse(is.na(.), 0, 1)) |>
as.data.frame()
upset_matrix = make_comb_mat(diabetes_studies_wide)
# create an upset plot
UpSet(upset_matrix,
comb_order = order(comb_size(upset_matrix), decreasing = TRUE)
#nsets = 3,
#order.by = "freq",
# sets.bar.color = "#56B4E9",
# main.bar.color = "#0072B2",
# matrix.color = "#D55E00",
#text.scale = c(1.5, 1.5, 1.5, 1.5, 1.5, 1.5)
)
Between different types of cancer
cancer_studies = disease_mapping_with_cause |>
filter(grepl("cancer|neoplasm", cause, ignore.case = TRUE)) |>
select(PUBMED_ID, cause) |>
distinct()
cancer_studies_wide = cancer_studies |>
tidyr::pivot_wider(names_from = cause, values_from = cause, values_fill = NA) |>
select(-PUBMED_ID) |>
mutate_all(~ ifelse(is.na(.), 0, 1)) |>
as.data.frame()
upset_matrix_cancer = make_comb_mat(cancer_studies_wide)
UpSet(upset_matrix_cancer,
comb_order = order(comb_size(upset_matrix_cancer), decreasing = TRUE)
#nsets = 3,
#order.by = "freq",
# sets.bar.color = "#56B4E9",
# main.bar.color = "#0072B2",
# matrix.color = "#D55E00",
#text.scale = c(1.5, 1.5, 1.5, 1.5, 1.5, 1.5)
)
Between different cardiovascular diseases
cardiovascular_studies = disease_mapping_with_cause |>
filter(grepl("heart|stroke|hemorrhage", cause, ignore.case = TRUE)) |>
select(PUBMED_ID, cause) |>
distinct()
cardiovascular_studies_wide = cardiovascular_studies |>
tidyr::pivot_wider(names_from = cause, values_from = cause, values_fill = NA) |>
select(-PUBMED_ID) |>
mutate_all(~ ifelse(is.na(.), 0, 1)) |>
as.data.frame()
upset_matrix_cardiovascular = make_comb_mat(cardiovascular_studies_wide)
UpSet(upset_matrix_cardiovascular,
comb_order = order(comb_size(upset_matrix_cardiovascular), decreasing = TRUE)
#nsets = 3,
#order.by = "freq",
# sets.bar.color = "#56B4E9",
# main.bar.color = "#0072B2",
# matrix.color = "#D55E00",
#text.scale = c(1.5, 1.5, 1.5, 1.5, 1.5, 1.5)
)
Across liver diseases
Cirrhosis and other chronic liver diseases, Liver cancer due to hepatitis C, Liver cancer due to hepatitis B, Diabetes
liver_studies = disease_mapping_with_cause |>
filter(grepl("liver|diabetes", cause, ignore.case = TRUE)) |>
filter(!grepl("Type 2|Type 1", cause, ignore.case = TRUE)) |>
select(PUBMED_ID, cause) |>
distinct()
liver_studies_wide = liver_studies |>
tidyr::pivot_wider(names_from = cause, values_from = cause, values_fill = NA) |>
select(-PUBMED_ID) |>
mutate_all(~ ifelse(is.na(.), 0, 1)) |>
as.data.frame()
upset_matrix_liver = make_comb_mat(liver_studies_wide)
UpSet(upset_matrix_liver,
comb_order = order(comb_size(upset_matrix_liver), decreasing = TRUE)
#nsets = 3,
#order.by = "freq",
# sets.bar.color = "#56B4E9",
# main.bar.color = "#0072B2",
# matrix.color = "#D55E00",
#text.scale = c(1.5, 1.5, 1.5, 1.5, 1.5, 1.5)
)
Across all non-communicable diseases
non_communicable_studies =
disease_mapping_with_cause |>
dplyr::filter(!cause %in% c("HIV/AIDS",
"Tuberculosis",
"Malaria",
"Lower respiratory infections",
"Diarrhoeal diseases",
"Neonatal disorders",
"Tetanus",
"Diphtheria",
"Pertussis" ,
"Measles",
"Maternal disorders")) |>
filter(!is.na(cause)) |>
select(PUBMED_ID, cause) |>
distinct()
non_communicable_studies_wide = non_communicable_studies |>
tidyr::pivot_wider(names_from = cause, values_from = cause, values_fill = NA) |>
select(-PUBMED_ID) |>
mutate_all(~ ifelse(is.na(.), 0, 1)) |>
as.data.frame()
upset_matrix_non_communicable = make_comb_mat(non_communicable_studies_wide)
UpSet(upset_matrix_non_communicable,
comb_order = order(comb_size(upset_matrix_non_communicable), decreasing = TRUE)
#nsets = 3,
#order.by = "freq",
# sets.bar.color = "#56B4E9",
# main.bar.color = "#0072B2",
# matrix.color = "#D55E00",
#text.scale = c(1.5, 1.5, 1.5, 1.5, 1.5, 1.5)
)
Between different respiratory diseases
Chronic obstructive pulmonary disease, “Tracheal, bronchus, and lung cancer
respiratory_studies = disease_mapping_with_cause |>
filter(grepl("chronic obstructive pulmonary disease|tracheal|bronchus|lung cancer", cause, ignore.case = TRUE)) |>
select(PUBMED_ID, cause) |>
distinct()
respiratory_studies_wide = respiratory_studies |>
tidyr::pivot_wider(names_from = cause, values_from = cause, values_fill = NA) |>
select(-PUBMED_ID) |>
mutate_all(~ ifelse(is.na(.), 0, 1)) |>
as.data.frame()
upset_matrix_respiratory = make_comb_mat(respiratory_studies_wide)
UpSet(upset_matrix_respiratory,
comb_order = order(comb_size(upset_matrix_respiratory), decreasing = TRUE)
#nsets = 3,
#order.by = "freq",
# sets.bar.color = "#56B4E9",
# main.bar.color = "#0072B2",
# matrix.color = "#D55E00",
#text.scale = c(1.5, 1.5, 1.5, 1.5, 1.5, 1.5)
)
Check recall i.e. that I have identified most/all studies of the priority diseases
Diabetes
no_capture <- paste0("(?<!no )",
"(?<!non )",
"(?<!no type 2 )"
)
diabetes_grep = paste0(no_capture, "diabetes")
disease_mapping_with_cause |>
filter(grepl(diabetes_grep,
`DISEASE/TRAIT`,
ignore.case = TRUE,
perl = TRUE)|
grepl(diabetes_grep,
collected_all_disease_terms,
ignore.case = TRUE,
perl = TRUE
)
) |>
group_by(PUBMED_ID) |>
filter(all(is.na(cause))) |>
select(PUBMED_ID,
`DISEASE/TRAIT`,
collected_all_disease_terms,
icd10_code,
cause) |>
distinct() # A tibble: 0 × 5
# Groups: PUBMED_ID [0]
# ℹ 5 variables: PUBMED_ID <int>, DISEASE/TRAIT <chr>,
# collected_all_disease_terms <chr>, icd10_code <chr>, cause <chr>Liver cancer, and other chronic liver diseases (e.g. cirrhosis)
Check recall for these together as they are often comorbid and may be captured by similar terms
liver_cancer_grep = paste0(no_capture, "liver|cirhosis|hepatocellular carcinoma|hepatoma|hepatic carcinoma")
disease_mapping_with_cause |>
filter(grepl(liver_cancer_grep,
`DISEASE/TRAIT`,
ignore.case = TRUE,
perl = TRUE)|
grepl(liver_cancer_grep,
collected_all_disease_terms,
ignore.case = TRUE,
perl = TRUE
)
) |>
group_by(PUBMED_ID) |>
filter(all(is.na(cause))) |>
group_by(PUBMED_ID,
`DISEASE/TRAIT`,
collected_all_disease_terms,
cause) |>
summarise(icd10_code = str_flatten(unique(icd10_code),
collapse = ", ",
na.rm = T),
cause = str_flatten(unique(icd10_code),
collapse = ", ",
na.rm = T)) |>
ungroup() |>
select(`DISEASE/TRAIT`,
collected_all_disease_terms) |>
distinct()`summarise()` has grouped output by 'PUBMED_ID', 'DISEASE/TRAIT',
'collected_all_disease_terms'. You can override using the `.groups` argument.# A tibble: 21 × 2
`DISEASE/TRAIT` collected_all_diseas…¹
<chr> <chr>
1 immunoglobulin light chain (al) amyloidosis (liver in… amyloidosis
2 biliverdin reductase a level in chronic kidney diseas… chronic kidney disease
3 biliverdin reductase a level in chronic kidney diseas… hypertension
4 fatty acid-binding protein, liver level in chronic ki… chronic kidney disease
5 fatty acid-binding protein, liver level in chronic ki… hypertension
6 glycogen phosphorylase, liver form level in chronic k… chronic kidney disease
7 glycogen phosphorylase, liver form level in chronic k… hypertension
8 hepatoma-derived growth factor level in chronic kidne… chronic kidney disease
9 hepatoma-derived growth factor level in chronic kidne… hypertension
10 hepatoma-derived growth factor-like protein 1 level i… chronic kidney disease
# ℹ 11 more rows
# ℹ abbreviated name: ¹collected_all_disease_termsStroke
stroke_grep = paste0(no_capture,
"stroke|cerebrovascular|cerebral|hemorrhage")
disease_mapping_with_cause |>
filter(grepl(stroke_grep,
`DISEASE/TRAIT`,
ignore.case = TRUE,
perl = TRUE)|
grepl(stroke_grep,
collected_all_disease_terms,
ignore.case = TRUE,
perl = TRUE
)
) |>
group_by(PUBMED_ID) |>
filter(all(is.na(cause))) |>
group_by(PUBMED_ID,
`DISEASE/TRAIT`,
collected_all_disease_terms,
cause) |>
summarise(icd10_code = str_flatten(unique(icd10_code),
collapse = ", ",
na.rm = T),
cause = str_flatten(unique(icd10_code),
collapse = ", ",
na.rm = T)) |>
ungroup() |>
select(`DISEASE/TRAIT`,
collected_all_disease_terms) |>
distinct()`summarise()` has grouped output by 'PUBMED_ID', 'DISEASE/TRAIT',
'collected_all_disease_terms'. You can override using the `.groups` argument.# A tibble: 30 × 2
`DISEASE/TRAIT` collected_all_diseas…¹
<chr> <chr>
1 progressive supranuclear palsy dementia with cerebra…
2 hippocampal atrophy other cerebral degene…
3 brain arteriovenous malformations (sporadic) cerebral arteriovenou…
4 cerebral amyloid angiopathy cerebral amyloid angi…
5 neuropathologic traits cerebral amyloid angi…
6 progressive supranuclear palsy and immune-mediated di… dementia with cerebra…
7 neuritic plaques or cerebral amyloid angiopathy cerebral amyloid angi…
8 neuritic plaques or neurofibrillary tangles or cerebr… cerebral amyloid angi…
9 neurofibrillary tangles or cerebral amyloid angiopathy cerebral amyloid angi…
10 non-richardson's syndrome vs richardson's syndrome in… dementia with cerebra…
# ℹ 20 more rows
# ℹ abbreviated name: ¹collected_all_disease_termsStomach cancer
stomach_cancer_grep = paste0(no_capture,
"stomach|gastric")
disease_mapping_with_cause |>
filter(grepl(stomach_cancer_grep,
`DISEASE/TRAIT`,
ignore.case = TRUE,
perl = TRUE)|
grepl(stomach_cancer_grep,
collected_all_disease_terms,
ignore.case = TRUE,
perl = TRUE
)
) |>
group_by(PUBMED_ID) |>
filter(all(is.na(cause))) |>
group_by(PUBMED_ID,
`DISEASE/TRAIT`,
collected_all_disease_terms,
cause) |>
summarise(icd10_code = str_flatten(unique(icd10_code),
collapse = ", ",
na.rm = T),
cause = str_flatten(unique(icd10_code),
collapse = ", ",
na.rm = T)) |>
ungroup() |>
select(`DISEASE/TRAIT`,
collected_all_disease_terms) |>
distinct()`summarise()` has grouped output by 'PUBMED_ID', 'DISEASE/TRAIT',
'collected_all_disease_terms'. You can override using the `.groups` argument.# A tibble: 15 × 2
`DISEASE/TRAIT` collected_all_diseas…¹
<chr> <chr>
1 weight loss (gastric bypass surgery) body weight loss
2 gastric atrophy atrophic gastritis
3 severe gastric atrophy atrophic gastritis
4 gastric or stomach ulcers gastric ulcer
5 gastroesophageal reflux disease or gastric reflux gastroesophageal refl…
6 stomach disorder stomach disease
7 gastric inhibitory polypeptide level in chronic kidne… chronic kidney disease
8 gastric inhibitory polypeptide level in chronic kidne… hypertension
9 gastric intrinsic factor level in chronic kidney dise… chronic kidney disease
10 gastric intrinsic factor level in chronic kidney dise… hypertension
11 chronic stomach pain abdominal pain
12 comorbidity of gastric and duodenal ulcers duodenal ulcer
13 comorbidity of gastric and duodenal ulcers gastric ulcer
14 gastric ulcer gastric ulcer
15 gastric reflux gastroesophageal refl…
# ℹ abbreviated name: ¹collected_all_disease_termsCervical cancer
cervical_cancer_grep = paste0(no_capture,
"cervical|cervix")
disease_mapping_with_cause |>
filter(grepl(cervical_cancer_grep,
`DISEASE/TRAIT`,
ignore.case = TRUE,
perl = TRUE)|
grepl(cervical_cancer_grep,
collected_all_disease_terms,
ignore.case = TRUE,
perl = TRUE
)
) |>
group_by(PUBMED_ID) |>
filter(all(is.na(cause))) |>
group_by(PUBMED_ID,
`DISEASE/TRAIT`,
collected_all_disease_terms,
cause) |>
summarise(icd10_code = str_flatten(unique(icd10_code),
collapse = ", ",
na.rm = T),
cause = str_flatten(unique(icd10_code),
collapse = ", ",
na.rm = T)) |>
ungroup() |>
select(`DISEASE/TRAIT`,
collected_all_disease_terms) |>
distinct()`summarise()` has grouped output by 'PUBMED_ID', 'DISEASE/TRAIT',
'collected_all_disease_terms'. You can override using the `.groups` argument.# A tibble: 8 × 2
`DISEASE/TRAIT` collected_all_diseas…¹
<chr> <chr>
1 cervical high-risk human papilloma virus infection human papilloma virus…
2 cervical high-risk human papilloma virus infection (pe… human papilloma virus…
3 cervical spondylosis cervical spondylosis
4 back problem (firth correction) abnormality of the ce…
5 back problem (spa correction) abnormality of the ce…
6 cervical dystonia cervical dystonia
7 uterine cancer cervical cancer
8 back problem abnormality of the ce…
# ℹ abbreviated name: ¹collected_all_disease_terms# pubmed 35729236 studies
# hormone-sensitive cancers: breast, ovarian, thyroid, prostate, and uterine cancer
# but GWAS Catalog mapped uterine cancer to cervical cancer, which is
# why it is picked up in this recall, but correctly not
# mapped to cervical cancer in ICD-10 mapping!
# human papilloma virus (HPV) is a major cause of cervical cancer ...
# so do we include studies of HPV in our mapping of cervical cancer?
hpv_grep <- paste0(no_capture,
"hpv|human papilloma virus|pap smear")
disease_mapping_with_cause |>
filter(grepl(hpv_grep,
`DISEASE/TRAIT`,
ignore.case = TRUE,
perl = TRUE)|
grepl(hpv_grep,
collected_all_disease_terms,
ignore.case = TRUE,
perl = TRUE
)
) |>
group_by(PUBMED_ID) |>
filter(all(is.na(cause))) |>
group_by(PUBMED_ID,
`DISEASE/TRAIT`,
collected_all_disease_terms,
cause) |>
summarise(icd10_code = str_flatten(unique(icd10_code),
collapse = ", ",
na.rm = T),
cause = str_flatten(unique(icd10_code),
collapse = ", ",
na.rm = T)) |>
ungroup() |>
select(`DISEASE/TRAIT`,
collected_all_disease_terms) |>
distinct()`summarise()` has grouped output by 'PUBMED_ID', 'DISEASE/TRAIT',
'collected_all_disease_terms'. You can override using the `.groups` argument.# A tibble: 12 × 2
`DISEASE/TRAIT` collected_all_diseas…¹
<chr> <chr>
1 hpv seropositivity hpv seropositivity
2 abnormal papanicolaou smear of cervix and cervical hpv abnormal pap smear
3 abnormal papanicolaou smear of cervix and cervical hpv hpv seropositivity
4 viral warts & hpv benign neoplasm
5 viral warts & hpv hpv seropositivity
6 cervical high-risk human papilloma virus infection human papilloma virus…
7 cervical high-risk human papilloma virus infection (p… human papilloma virus…
8 abnormal smear cervix abnormal pap smear
9 icd10 r87619: abnormal smear cervix abnormal pap smear
10 human papillomavirus seropositivity (hpv16 e6/e7/l1 o… hpv seropositivity
11 human papillomavirus seropositivity (hpv16 e6/e7/l1) hpv seropositivity
12 human papillomavirus seropositivity (hpv18 l1) hpv seropositivity
# ℹ abbreviated name: ¹collected_all_disease_termsdisease_mapping_with_cause <-
disease_mapping_with_cause |>
mutate(icd10_code = str_replace(icd10_code, "R879", "R87.9")) Lip and oral cavity cancer
lip_oral_cavity_cancer_grep = paste0(no_capture,
"lip\\b|oral cavity|mouth|tongue|pharynx|larynx|pharyngeal|laryngeal")
disease_mapping_with_cause |>
filter(grepl(lip_oral_cavity_cancer_grep,
`DISEASE/TRAIT`,
ignore.case = TRUE,
perl = TRUE)|
grepl(lip_oral_cavity_cancer_grep,
collected_all_disease_terms,
ignore.case = TRUE,
perl = TRUE
)
) |>
filter(!grepl("cleft", `DISEASE/TRAIT`, ignore.case = TRUE)) |>
group_by(PUBMED_ID) |>
filter(all(is.na(cause))) |>
group_by(PUBMED_ID,
`DISEASE/TRAIT`,
collected_all_disease_terms,
cause) |>
summarise(icd10_code = str_flatten(unique(icd10_code),
collapse = ", ",
na.rm = T),
cause = str_flatten(unique(icd10_code),
collapse = ", ",
na.rm = T))`summarise()` has grouped output by 'PUBMED_ID', 'DISEASE/TRAIT',
'collected_all_disease_terms'. You can override using the `.groups` argument.# A tibble: 22 × 5
# Groups: PUBMED_ID, DISEASE/TRAIT, collected_all_disease_terms [22]
PUBMED_ID `DISEASE/TRAIT` collected_all_diseas…¹ cause icd10_code
<int> <chr> <chr> <chr> <chr>
1 29855589 velopharyngeal dysfunction velopharyngeal dysfun… Q35,… Q35, Q38.8
2 30837455 mouth ulcers oral ulcer K12.0 K12.0
3 31235808 mouth ulcers oral ulcer K12.0 K12.0
4 32634583 cv-a6-associated hand, foo… hand foot and mouth d… B08.4 B08.4
5 33959723 throat or larynx disorder laryngeal disease J38.… J38.7, S1…
6 33959723 throat or larynx disorder throat disease R04.… R04.0, R0…
7 34662886 icd10 d10: benign neoplasm… benign neoplasm D10 D10
8 34662886 icd10 d37: neoplasm of unc… neoplasm D37 D37
9 34662886 icd10 j38.7: other disease… laryngeal disease J38.7 J38.7
10 34662886 icd10 j38: diseases of voc… laryngeal disease J38 J38
# ℹ 12 more rows
# ℹ abbreviated name: ¹collected_all_disease_terms# !!!!!!!!!!!! Maybe caught one! Heart disease
heart_disease_grep = paste0(no_capture,
"heart disease|cardiovascular|coronary|myocardial|angina|heart failure")
disease_mapping_with_cause |>
filter(grepl(heart_disease_grep,
`DISEASE/TRAIT`,
ignore.case = TRUE,
perl = TRUE)|
grepl(heart_disease_grep,
collected_all_disease_terms,
ignore.case = TRUE,
perl = TRUE
)
) |>
group_by(PUBMED_ID) |>
filter(all(is.na(cause))) |>
group_by(PUBMED_ID,
`DISEASE/TRAIT`,
collected_all_disease_terms,
cause) |>
summarise(icd10_code = str_flatten(unique(icd10_code),
collapse = ", ",
na.rm = T),
cause = str_flatten(unique(icd10_code),
collapse = ", ",
na.rm = T))`summarise()` has grouped output by 'PUBMED_ID', 'DISEASE/TRAIT',
'collected_all_disease_terms'. You can override using the `.groups` argument.# A tibble: 90 × 5
# Groups: PUBMED_ID, DISEASE/TRAIT, collected_all_disease_terms [90]
PUBMED_ID `DISEASE/TRAIT` collected_all_diseas…¹ cause icd10_code
<int> <chr> <chr> <chr> <chr>
1 17903304 heart failure heart failure I50,… I50, I50.…
2 17903304 major cvd cardiovascular disease I51.6 I51.6
3 20400778 mortality in heart failure heart failure I50 I50
4 20445134 heart failure heart failure I50,… I50, I50.…
5 20838585 cardiovascular risk factors cardiovascular disease I51.6 I51.6
6 20838585 cardiovascular risk factor… cardiovascular disease I51.6 I51.6
7 21779381 cardiovascular disease ris… cardiovascular disease I51.6 I51.6
8 23297363 tetralogy of fallot congenital anomaly of… Q21.3 Q21.3
9 23337848 vascular constriction cardiovascular disease I51.6 I51.6
10 23708190 congenital heart malformat… congenital anomaly of… I35.… I35.0, Q2…
# ℹ 80 more rows
# ℹ abbreviated name: ¹collected_all_disease_termsCheck precision i.e. that the studies I have mapped to the priority diseases are actually about those diseases
Type 2 diabetes
disease_mapping_with_cause |>
filter(cause == "Type 2 diabetes mellitus") |>
filter(!grepl("type 2 diabetes|diabetes type 2|e11|type ii diabetes", `DISEASE/TRAIT`, ignore.case = TRUE)
) |>
group_by(PUBMED_ID,
STUDY_ACCESSION,
`DISEASE/TRAIT`,
collected_all_disease_terms) |>
summarise(icd10_code = str_flatten(unique(icd10_code),
collapse = ", ",
na.rm = T),
cause = str_flatten(unique(cause),
collapse = ", ",
na.rm = T)) |>
select(PUBMED_ID,
`DISEASE/TRAIT`,
collected_all_disease_terms,
icd10_code,
cause) |>
distinct()`summarise()` has grouped output by 'PUBMED_ID', 'STUDY_ACCESSION',
'DISEASE/TRAIT'. You can override using the `.groups` argument.
Adding missing grouping variables: `STUDY_ACCESSION`# A tibble: 134 × 6
# Groups: PUBMED_ID, STUDY_ACCESSION, DISEASE/TRAIT [134]
STUDY_ACCESSION PUBMED_ID `DISEASE/TRAIT` collected_all_diseas…¹ icd10_code
<chr> <int> <chr> <chr> <chr>
1 GCST000108 17903298 diabetes (incide… diabetes mellitus E11
2 GCST001965 23575436 glycemic traits type 2 diabetes melli… E11
3 GCST002025 23670970 cystic fibrosis-… type 2 diabetes melli… E11
4 GCST009161 23696881 diabetes biomark… diabetes mellitus E11
5 GCST002153 23982368 cardiovascular h… type 2 diabetes melli… E11
6 GCST002637 25317111 medication adher… diabetes mellitus E11
7 GCST003250 26631737 urinary albumin-… diabetes mellitus E11
8 GCST003373 26831199 glomerular filtr… diabetes mellitus E11
9 GCST003457 26915486 soluble receptor… type 2 diabetes melli… E11
10 GCST003797 27670767 diabetes in resp… diabetes mellitus E11
# ℹ 124 more rows
# ℹ abbreviated name: ¹collected_all_disease_terms
# ℹ 1 more variable: cause <chr># problems with GCST90079743, GCST90083729 study provided mapping
# pubmed id 29703844 - should map to type 2 diabetes
# genome-wide association study (GWAS) of DKD in type 2 diabetes (T2D)
# ? idk: 27670767Checking hierarchy levels
# disease_mapping_with_cause =
# disease_mapping_with_cause |>
# filter(!is.na(cause))
# filtering na to deal with pivoting
# disease_mapping_with_cause_grouped =
# disease_mapping_with_cause |>
# select(-icd10_code_num) |>
# tidyr::pivot_wider(
# id_cols = c("collected_all_disease_terms",
# "icd10_code",
# "icd10_description"),
# names_from = cause_hierarchy_level,
# names_glue = "l{.name}_{.value}",
# values_from = cause,
# values_fn = ~str_flatten(unique(.x), collapse = ", ", na.rm = TRUE)
# )# check which l2 causes are NA but l3 causes are not NA
# can use l3 causes therefore to fill in l2 causes
# disease_mapping_with_cause_grouped |>
# rowwise() |>
# filter(is.na(l2_cause) & !is.na(l3_cause)) |>
# head()
#
# neoplasms = c("Other neoplasms",
# "Hodgkin lymphoma",
# "Leukemia",
# "Non-Hodgkin lymphoma",
# "Multiple myeloma")
#
# disease_mapping_with_cause_grouped =
# disease_mapping_with_cause_grouped |>
# rowwise() |>
# mutate(l2_cause =
# list(ifelse(is.na(l2_cause),
# case_when(map_lgl(l3_cause, ~ .x %in% neoplasms) ~ "Neoplasms",
# map_lgl(l3_cause, ~ .x %in% "Neonatal disorders") ~ "Maternal and neonatal disorders",
# map_lgl(l3_cause, ~ .x %in% "Other cardiovascular and circulatory diseases") ~ "Cardiovascular diseases"
# ),
# l2_cause
# )
# )) |>
# ungroup()
# to_map_manually <-
# disease_mapping_with_cause_grouped |>
# rowwise() |>
# filter(is.null(l2_cause)) |>
# pull(collected_all_disease_terms) |>
# unique()
#
#
# update_unmapped_cause = function(df,
# unmapped_term,
# l2_cause_rep,
# l3_cause_rep){
#
# df =
# df |>
# mutate(l3_cause =
# ifelse(collected_all_disease_terms == unmapped_term,
# l3_cause_rep,
# l3_cause
# )) |>
# mutate(l2_cause =
# ifelse(collected_all_disease_terms == unmapped_term,
# l2_cause_rep,
# l2_cause
# ))
#
# return(df)
#
# }
#
# disease_mapping_with_cause_grouped =
# update_unmapped_cause(disease_mapping_with_cause_grouped,
# "bone cancer",
# "Neoplasms",
# "Malignant neoplasms of bone and articular cartilage")Save + combine with GWAS data
# disease_mapping_orig <- data.table::fread(
# here::here("output/icd_map/gwas_disease_to_icd10_mapping.csv")
# )
#
# disease_mapping_with_cause_grouped =
# disease_mapping_with_cause_grouped |>
# tidyr::unnest_longer(c("l3_cause", "l4_cause"))
disease_mapping_final = disease_mapping_with_cause
# left_join(
# disease_mapping_orig,
# disease_mapping_with_cause,
# by = c("collected_all_disease_terms",
# "icd10_code",
# "icd10_description")
# )
# gwas_study_info <- fread(here::here("output/gwas_cat/gwas_study_info_group_v2.csv"))
#
# gwas_study_info =
# gwas_study_info |>
# select(`DISEASE/TRAIT`, PUBMED_ID, STUDY_ACCESSION, COHORT)
#
# disease_mapping_final =
# left_join(
# disease_mapping_final,
# gwas_study_info
# )
# disease_mapping_final =
# disease_mapping_final |>
# group_by(STUDY_ACCESSION, collected_all_disease_terms) |>
# summarise(
# PUBMED_ID = paste0(unique(PUBMED_ID), collapse = "; "),
# COHORT = paste0(unique(COHORT), collapse = "; "),
# icd10_code = paste0(unique(icd10_code), collapse = "; "),
# icd10_description = paste0(unique(icd10_description), collapse = "; "),
# # l2_cause = str_flatten(unique(str_split(l2_cause, ", ")), # add separation step here ...
# # collapse = "; ", na.rm = TRUE),
# l3_cause = str_flatten(unique(str_split(l3_cause, ", ")),
# collapse = "; ", na.rm = TRUE),
# l4_cause = str_flatten(unique(str_split(l4_cause, ", ")),
# collapse = "; ", na.rm = TRUE)
# )
data.table::fwrite(
# disease_mapping_final,
disease_mapping_with_cause,
here::here("output/icd_map/gwas_study_gbd_causes.csv")
)
sessionInfo()R version 4.3.1 (2023-06-16)
Platform: aarch64-apple-darwin20 (64-bit)
Running under: macOS 26.3.1
Matrix products: default
BLAS: /Library/Frameworks/R.framework/Versions/4.3-arm64/Resources/lib/libRblas.0.dylib
LAPACK: /Library/Frameworks/R.framework/Versions/4.3-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.11.0
locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
time zone: America/Los_Angeles
tzcode source: internal
attached base packages:
[1] grid stats graphics grDevices datasets utils methods
[8] base
other attached packages:
[1] ComplexHeatmap_2.18.0 data.table_1.17.8 tidyr_1.3.1
[4] purrr_1.1.0 stringr_1.6.0 dplyr_1.1.4
[7] workflowr_1.7.2
loaded via a namespace (and not attached):
[1] shape_1.4.6.1 circlize_0.4.16 rjson_0.2.23
[4] xfun_0.55 bslib_0.9.0 GlobalOptions_0.1.2
[7] processx_3.8.6 callr_3.7.6 vctrs_0.6.5
[10] tools_4.3.1 ps_1.9.1 generics_0.1.4
[13] stats4_4.3.1 parallel_4.3.1 tibble_3.3.0
[16] cluster_2.1.4 pkgconfig_2.0.3 RColorBrewer_1.1-3
[19] S4Vectors_0.40.2 readxl_1.4.5 lifecycle_1.0.4
[22] compiler_4.3.1 git2r_0.36.2 getPass_0.2-4
[25] codetools_0.2-19 clue_0.3-66 httpuv_1.6.16
[28] htmltools_0.5.8.1 sass_0.4.10 yaml_2.3.10
[31] later_1.4.4 pillar_1.11.1 crayon_1.5.3
[34] jquerylib_0.1.4 whisker_0.4.1 cachem_1.1.0
[37] iterators_1.0.14 foreach_1.5.2 tidyselect_1.2.1
[40] digest_0.6.37 stringi_1.8.7 rprojroot_2.1.0
[43] fastmap_1.2.0 here_1.0.1 colorspace_2.1-2
[46] cli_3.6.5 magrittr_2.0.4 utf8_1.2.6
[49] withr_3.0.2 promises_1.3.3 rmarkdown_2.30
[52] httr_1.4.7 matrixStats_1.5.0 cellranger_1.1.0
[55] png_0.1-8 GetoptLong_1.0.5 evaluate_1.0.5
[58] knitr_1.50 IRanges_2.36.0 doParallel_1.0.17
[61] rlang_1.1.6 Rcpp_1.1.0 glue_1.8.0
[64] BiocManager_1.30.26 renv_1.1.8 BiocGenerics_0.48.1
[67] rstudioapi_0.17.1 jsonlite_2.0.0 R6_2.6.1
[70] fs_1.6.6