getXrefList gives strange result for systemCode "Ce"

[ayeTown]

Hi there,
Thanks for this great tool! I have been using it to grab metabolites associated with each wikipathway. When I use the getXrefList function to grab all CHEBI IDs associated with a pathway I get the following result (for WP106):

[1] "15351"       "15570"       "15682"       "15727"       "15741"       "15859"       "15919"       "16958"       "16977"      
[10] "17053"       "17196"       "17364"       "18012"       "30744"       "30769"       "30915"       "32816"       "33190"      
[19] "57288"       "57416"       "57485"       "57589"       "57735"       "57966"       "57972"       "58048"       "6650"       
[28] "CHEBI:15351" "CHEBI:15570" "CHEBI:15682" "CHEBI:15727" "CHEBI:15741" "CHEBI:15859" "CHEBI:15919" "CHEBI:16452" "CHEBI:16810"
[37] "CHEBI:16958" "CHEBI:16977" "CHEBI:17053" "CHEBI:17196" "CHEBI:17364" "CHEBI:18012" "CHEBI:29806" "CHEBI:30744" "CHEBI:30769"
[46] "CHEBI:30915" "CHEBI:32816" "CHEBI:33190" "CHEBI:57288" "CHEBI:57416" "CHEBI:57485" "CHEBI:57589" "CHEBI:57735" "CHEBI:57966"
[55] "CHEBI:57972" "CHEBI:58048" "CHEBI:6650" 

Why is the result duplicated? Should I just grab all IDs starting with CHEBI?

[khanspers]

Hi @ayeTown, I tried it with the same pathway and I don't get the duplicates, only identifiers without the CHEBI prefix. Which version of rWikiPathways are you using? I have 1.28.0.

[ayeTown]

@khanspers version 1.18.2. I also have R version 4.2.1 and get the message that the package is not available for my version of R when I try to re-install.

[ayeTown]

Installing the development version now with BiocManager. Will report back if this fixes the issue.

[ayeTown]

Updated R, R Studio, and BiocManager. Now I do get a list of just the IDs, but I get this strange warning message which I've never seen before. Any ideas?

[ayeTown]

Also, with my previous version I was able to grab 1576 total human pathways and now that number has dropped down to 976. Have some been removed from version 1.18.2 to 1.29.0?

[ayeTown]

Also, the systemCode "Lm" in the getXrefList to grab LIPID MAPS IDs no longer works :(

[egonw]

Why is the result duplicated? Should I just grab all IDs starting with CHEBI?

I would indeed pick those without CHEBI:.

@ayeTown, that's a bit historical and it's a bit early to remember the full history. It had to do with prefixes, the ontology using the CHEBI_xxxx pattern, IDs being literal strings or if the content can be interpreted, and people trying to just make it to work.

[egonw]

Have some been removed from version 1.18.2 to 1.29.0?

I think version 1.22 was where we switched from the classic website to the new website. I need to confirm this (@AlexanderPico?), but I think in version 1.22 the package started using the JSON from the new website. And the new website has fewer pathways than the old website. That could maybe explain it.

[ayeTown]

So should I only use the newer version of the package? Any idea why the pathways were removed from the new website? Although, I can't use the new package if I am working with LMIDs, due to the "Lm" systemCode no longer working.

[khanspers]

Have some been removed from version 1.18.2 to 1.29.0?

I think version 1.22 was where we switched from the classic website to the new website. I need to confirm this (@AlexanderPico?), but I think in version 1.22 the package started using the JSON from the new website. And the new website has fewer pathways than the old website. That could maybe explain it.

Confirmed. The switch to using JSON happened in version 1.22:

rWikiPathways/NEWS

Line 7 in 7f79176

+ Reimplementation of every function, replacing web service with static JSON

[khanspers]

Have some been removed from version 1.18.2 to 1.29.0?

I think version 1.22 was where we switched from the classic website to the new website. I need to confirm this (@AlexanderPico?), but I think in version 1.22 the package started using the JSON from the new website. And the new website has fewer pathways than the old website. That could maybe explain it.

I think that's right: the new official site does not include pathways converted from Reactome in the count for human pathways, which the classic wiki site did.

[ayeTown]

So the old site included pathways from Reactome? and the new site removed these? I'm just curious why they would remove them.

[khanspers]

Also, the systemCode "Lm" in the getXrefList to grab LIPID MAPS IDs no longer works :(

@egonw @AlexanderPico: It seems some system codes are no longer supported in getXrefList, for example "Lm" (LIPID MAPS). Error:

Must provide a supported systemCode, e.g., L (see docs)

Is this a bug or something that was dropped intentionally?