[BioC] Inconsistent results returned from KEGG

[michael watson (IAH-C)]

Hi
 
In my experience, the soap querying of KEGG has always been a bit flaky, whether using the perl modules or through the KEGGSOAP package.
 
Eg
 
> get.pathways.by.genes(c("eco:b0077", "eco:b0078"))
Type: string  & length =  4 
[1] "path:eco00290" "path:eco00650" "path:eco00660" "path:eco00770"
Warning messages:
1: Don't understand the SOAP type `string' yet in: fromSOAP(x, type = type, root = root, converters = converters,  
2: Don't understand the SOAP type `string' yet in: fromSOAP(x, type = type, root = root, converters = converters,  
3: Don't understand the SOAP type `string' yet in: fromSOAP(x, type = type, root = root, converters = converters,  
4: Don't understand the SOAP type `string' yet in: fromSOAP(x, type = type, root = root, converters = converters,  
 
> get.pathways.by.genes(c("eco:b0077"))
NULL
 
> get.pathways.by.genes(c("eco:b0078"))
NULL

The same is true for salmonella:

> query_vector[1:3]
[1] "stm:STM4113" "stm:STM2974" "stm:STM0515"
> get.pathways.by.genes(query_vector[1:2])
Type: string  & length =  1 
[1] "path:stm00051"
Warning message:
Don't understand the SOAP type `string' yet in: fromSOAP(x, type = type, root = root, converters = converters, 

> get.pathways.by.genes(query_vector[1:3])
NULL


To summarise:

1) get.pathways.by.genes returns warning messages

2) get.pathways.by.genes returns an incorrect result when given a vector of genes of length 1

3) get.pathways.by.genes returns inconsistent results when given vectors of different sizes, one of which contains genes that do not exist in a KEGG pathway

I've had similar issues with the perl access to KEGG and raised this with the helpdesk years ago.

Does anyone else have these issues?

Thanks

Mick