[BUG] Mapping ChEMBL IDs to ChEBI IDs using the 16/04/2024 metabolite bridge file produces inconsistent duplicate ChEBI IDs

[pklemmer]

Describe the bug

Using maps() to map ChEMBL IDs from an input df like:

source identifier
Cl / CHEMBL1091
Cl / CHEMBL11
Cl / CHEMBL99

to ChEBI ID using the metabolites20240416.bridge file as loadDatabase() argument produces inconsistently mapped duplicate ChEBI IDs:

source identifier target mapping isPrimary
Cl / CHEMBL1091 / Ce / CHEBI:17609 / T
Cl / CHEMBL1091 / Ce / 17609 / F

but also both duplicate IDs being indicated as primary:

source identifier target mapping isPrimary
Cl / CHEMBL11 / Ce / CHEBI:47499 / T
Cl / CHEMBL11 / Ce / 47499 / T

or even duplicate IDs being indicated as both true and false primary IDs:

source identifier target mapping isPrimary
Cl / CHEMBL1152 / Ce / CHEBI:8380 / T
Cl / CHEMBL1152 / Ce / 8380 / F
Cl / CHEMBL1152 / Ce / 8380 / T

Provide a minimally reproducible example (reprex)

The 'identifiers' argument for the maps() function is an input dataframe such as:

source identifier
Cl / CHEMBL1091
Cl / CHEMBL11
Cl / CHEMBL99

which was generated like this:

metabolite_input <- data.frame(
source = rep("Cl", length(mapped_chembls[, 1])),
identifier = mapped_chembls[, 1]
)

where mapped_chembls is a data frame with a single column containing one CHEMBL ID in the format 'CHEMBL123' per row.

The 'mapper' argument is an absolute file path like:

"C:/Users/user/Documents/GitHub/repo/BridgeDb/metabolites_20240416.bridge"

and the 'target' argument is 'Ce' to map to ChEBI.

Expected behavior

I believe that ChEBI IDs are typically associated with single unique ChEMBL IDs, so an ideal output should look like:

source identifier target mapping isPrimary
Cl / CHEMBL1152 / Ce / CHEBI:8380 / T

With the "CHEBI:" prefix in front of the actual ID.

R Session Information

Please report the output of either sessionInfo() or
sessioninfo::session_info() here.

options(width = 120)
R version 4.3.3 (2024-02-29 ucrt)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 10 x64 (build 19045)

Matrix products: default


locale:
[1] LC_COLLATE=English_Europe.utf8  LC_CTYPE=English_Europe.utf8    LC_MONETARY=English_Europe.utf8 LC_NUMERIC=C                    LC_TIME=English_Europe.utf8    

time zone: Europe/Berlin
tzcode source: internal

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
 [1] reprex_2.1.0         curl_5.2.1           BridgeDbR_2.10.2     rJava_1.0-11         RCy3_2.20.2          rWikiPathways_1.20.0 tidyr_1.3.1          rvest_1.0.4         
 [9] gprofiler2_0.2.3     stringr_1.5.1        httr_1.4.7           dplyr_1.1.4         

loaded via a namespace (and not attached):
 [1] gtable_0.3.4        rjson_0.2.21        ggplot2_3.5.0       htmlwidgets_1.6.4   caTools_1.18.2      vctrs_0.6.5         tools_4.3.3         bitops_1.0-7       
 [9] generics_0.1.3      stats4_4.3.3        base64url_1.4       tibble_3.2.1        fansi_1.0.6         pkgconfig_2.0.3     KernSmooth_2.23-22  data.table_1.15.4  
[17] RColorBrewer_1.1-3  uuid_1.2-0          graph_1.78.0        lifecycle_1.0.4     compiler_4.3.3      gplots_3.1.3.1      munsell_0.5.1       repr_1.1.7         
[25] uchardet_1.1.1      htmltools_0.5.8.1   RCurl_1.98-1.14     lazyeval_0.2.2      plotly_4.10.4       pillar_1.9.0        crayon_1.5.2        gtools_3.9.5       
[33] tidyselect_1.2.1    digest_0.6.35       stringi_1.8.3       purrr_1.0.2         RJSONIO_1.3-1.9     fastmap_1.1.1       grid_4.3.3          colorspace_2.1-0   
[41] cli_3.6.2           magrittr_2.0.3      base64enc_0.1-3     XML_3.99-0.16.1     utf8_1.2.4          IRdisplay_1.1       withr_3.0.0         scales_1.3.0       
[49] backports_1.4.1     IRkernel_1.3.2      pbdZMQ_0.3-10       evaluate_0.23       viridisLite_0.4.2   rlang_1.1.3         glue_1.7.0          selectr_0.4-2      
[57] BiocManager_1.30.22 xml2_1.3.6          BiocGenerics_0.46.0 pkgload_1.3.4       rstudioapi_0.16.0   jsonlite_1.8.8      R6_2.5.1            fs_1.6.3           

Indicate whether BiocManager::valid() returns TRUE.

BiocManager::valid() returns
"4 packages out-of-date; 0 packages too new"

Is the package installed via bioconda?

BridgeDbR is installed via BiocManager.

[egonw]

Thanks for filing the issue! I need to get some details together. The problem is probably in the ID mapping file, and therefore caused by how we create it, hence the transfer.