CTS retrieving non-live CIDs #225

schymane · 2019-12-13T11:19:19Z

We should make our default source of CIDs PubChem, and not CTS. There are too many discrepancies/error cropping up.
@meier-rene we may have to check the "status" of CIDs during validation, to catch and fix.

Example from freshly-created infolist:
https://pubchem.ncbi.nlm.nih.gov/compound/4644

tsufz · 2019-12-13T13:07:55Z

Hi Emma, This is a great enhancement. Using the source data source is always the best option. Have a nice weekend, Tobi

schymane · 2020-04-16T11:47:03Z

getPcId will also return non-live CIDs for non-standard tautomeric forms. We can fix this using a function in RChemMass, but we should make sure we upgrade the getPcId function to automatically do this. On the "todo" list ...

> getPcId("NPZTUJOABDZTLV-UHFFFAOYSA-N")
[1] 2759291
> getPCIDs.CIDtype(getPcId("NPZTUJOABDZTLV-UHFFFAOYSA-N"),type="preferred")
[1] 135399369

schymane · 2020-04-17T06:06:59Z

@meowcat I will need to upgrade getPcId to make sure it returns "live" CIDs, this is fine - but where can I check whether we grab PubChem CIDs from PubChem vs CTS? If we already use getPcId (not CTS), then all I will need to do is fix that function, then this issue is solved. Thanks.

meowcat · 2020-04-17T08:28:36Z

So:
https://github.com/MassBank/RMassBank/createMassBank.R
Line 577 is where we call gatherPubChem to get data off PubChem.

RMassBank/R/createMassBank.R

Line 577 in 611b785

PcInfo <- gatherPubChem(inchikey_split)

Lines 602..608 is where we get CTS data. To do: what do we still need from CTS at this point?

RMassBank/R/createMassBank.R

Lines 602 to 608 in 611b785

    
           ##Use CTS to retrieve information 
        
           CTSinfo <- getCtsRecord(inchikey_split) 
        
           if((CTSinfo[1] == "Sorry, we couldn't find any matching results") || is.null(CTSinfo[1])) 
        
           { 
        
           	CTSinfo <- NA 
        
           }

Lines 775-786 is where we decide which PubChem ID to use. I guess you want to drop CTS completely as an option?

RMassBank/R/createMassBank.R

Lines 775 to 786 in 611b785

    
           # PubChem CID 
        
           if(is.na(PcInfo$PcID[1])){ 
        
           	if(!is.na(CTSinfo[1])){ 
        
           		if("PubChem CID" %in% CTS.externalIdTypes(CTSinfo)) 
        
           		{ 
        
           			pc <- CTS.externalIdSubset(CTSinfo,"PubChem CID") 
        
           			link[["PUBCHEM"]] <- paste0(min(pc)) 
        
           		} 
        
           	} 
        
           } else{ 
        
           	link[["PUBCHEM"]] <- PcInfo$PcID[1] 
        
           }

Then the actual data retrieval from PubChem is in gatherPubChem, where getPcId is called:

RMassBank/R/createMassBank.R

Lines 454 to 466 in 611b785

    
           gatherPubChem <- function(key){ 
        
           	PubChemData <- list() 
        
           	##Trycatches are there because pubchem has connection issues 1 in 50 times. 
        
           	##Write NA into the respective fields if something goes wrong with the conenction or the data. 
        
           	##Retrieve Pubchem CID 
        
           	tryCatch( 
        
           		PubChemData$PcID <- getPcId(key), 
        
           		error=function(e){ 
        
           		PubChemData$PcID <<- NA 
        
           	})

getPcId is then the function in webAccess,R:

RMassBank/R/webAccess.R

Lines 109 to 144 in 611b785

    
           getPcId <- function(query, from = "inchikey") 
        
           { 
        
           	baseURL <- "https://pubchem.ncbi.nlm.nih.gov/rest/pug/compound" 
        
           	url <- paste(baseURL, from, query, "description", "json", sep="/") 
        
           	errorvar <- 0 
        
           	currEnvir <- environment() 
        
           	tryCatch( 
        
           		data <- getURL(URLencode(url),timeout=5), 
        
           		error=function(e){ 
        
           		currEnvir$errorvar <- 1 
        
           	}) 
        
           	if(errorvar){ 
        
           		return(NA) 
        
           	} 
        
           	# This happens if the InChI key is not found: 
        
           	r <- fromJSON(data) 
        
           	if(!is.null(r$Fault)) 
        
           	return(NA) 
        
           	titleEntry <- which(unlist(lapply(r$InformationList$Information, function(i) !is.null(i$Title)))) 
        
           	titleEntry <- titleEntry[which.min(sapply(titleEntry, function(x)r$InformationList$Information[[x]]$CID))] 
        
           	PcID <- r$InformationList$Information[[titleEntry]]$CID 
        
           	if(is.null(PcID)){ 
        
           		return(NA) 
        
           	} else{ 
        
           		return(PcID) 
        
           	} 
        
           }

schymane · 2020-04-18T15:14:17Z

@meowcat I've created a new branch:
https://github.com/MassBank/RMassBank/tree/preferredPCIDs

I've added getPCIDs.CIDtype, adjusted getPcId and createMassBank.R (and updated my old email address). I'm stuck on the documentation - see emails.

schymane added bug enhancement labels Dec 13, 2019

MassBank / RMassBank

CTS retrieving non-live CIDs #225

CTS retrieving non-live CIDs #225

schymane commented Dec 13, 2019

This comment has been minimized.

tsufz commented Dec 13, 2019

This comment has been minimized.

schymane commented Apr 16, 2020

This comment has been minimized.

schymane commented Apr 17, 2020

This comment has been minimized.

meowcat commented Apr 17, 2020

This comment has been minimized.

schymane commented Apr 18, 2020

MassBank / RMassBank

Join GitHub today

CTS retrieving non-live CIDs #225

CTS retrieving non-live CIDs #225

Comments

schymane commented Dec 13, 2019

This comment has been minimized.

tsufz commented Dec 13, 2019

This comment has been minimized.

schymane commented Apr 16, 2020

This comment has been minimized.

schymane commented Apr 17, 2020

This comment has been minimized.

meowcat commented Apr 17, 2020

This comment has been minimized.

schymane commented Apr 18, 2020