GOseq errors

[egorleg]

Hi

Below is the code I used and the errors returned for the GOseq R package. I would appreciate any suggestions as to how to resolve them.

Regards

Bill

________________________________________________________

> library(goseq)
> de.genes <- scan("DE_genes.txt", what=character() )
Read 2744 items
> all.genes <- scan("all_genes.txt", what=character() )
Read 35180 items
> genes = as.integer(all.genes %in% de.genes)
> names(genes) = all.genes
> pwf = nullp(genes, "mm9", "ensGene")
Loading mm9 length data...
> mapping=read.table("innatedb.in",header=TRUE,sep="\t")
> innatedb=split(mapping$Pathway,mapping$ID)
> pathways=goseq(genes,pwf,gene2cat=innatedb)
Using manually entered categories.
Error in summary(map)[, 1] : incorrect number of dimensions
> head(pathways, n = 50)
Error in head(pathways, n = 50) : object 'pathways' not found
> enriched.pathways = pathways$category[pathways$upval < 0.01]
Error: object 'pathways' not found
> enriched.pathways
Error: object 'enriched.pathways' not found

[colinm]

I ran into the same problem using manual annotations. If you look in the code, the goseq function restricts the list gene2cat to names that are present in the rownames of the data.frame kicked out by nullp. If you don't add the names of the data used to generate the list to the rownames of the data.frame, it chokes in the 'reversemapping' function because it is dealing with a null list. This fixed the problem for me:

N = nullp(DEgenes = DETECTED_VECTOR,bias.data=LENGTHS_VECTOR)
rownames(N) <- names(LENGTHS_VECTOR) ##this is what fixes it
go = goseq(N,gene2cat=GENES_TO_CAT_LIST,method='Wallenius')

Obviously you'll have to tailor it a bit to your situation, but I hope that helps.

-Colin

[crh]

Hi,

I'm running into the same problem I think.

> sessionInfo()
R version 2.12.2 (2011-02-25)
Platform: i386-apple-darwin9.8.0/i386 (32-bit)

other attached packages:
[1] goseq_1.2.0 geneLenDataBase_0.99.5 BiasedUrn_1.03


Here's a snippet of my code:

> all <- read.delim("qd_de_gene_size_down_only.txt",header=T,stringsAsFactors=FALSE)
> head(all)
gene CDS DE
1 au5.g1.t1 927 0
2 au5.g10.t1 795 0

>
> #list of assayed genes
> assayed.genes<-all$gene
> head(assayed.genes)
[1] "au5.g1.t1" "au5.g10.t1" "au5.g100.t1" "au5.g1000.t1"
[5] "au5.g10000.t1" "au5.g10001.t1"
> is.vector(assayed.genes)
[1] TRUE
>
> #read gene lengths (CDS)
> gene.length<-all$CDS
> head(gene.length)
[1] 927 795 1941 2317 207 4239
> is.vector(gene.length)
[1] TRUE
>
> #list of DE genes (0|1) caution: use UP_ONLY file for goseq, or DOWN_only
> de.genes<-all$DE
> head(de.genes)
[1] 0 0 0 0 0 0
> is.vector(de.genes)
[1] TRUE
>
>
> #gene vector
> gene.vector<-as.integer(de.genes)
> names(gene.vector)<-assayed.genes
>
> #KEGG annotations data frame
> #au5.g10000_t1 520259 Ribosome
> go<-read.delim("../ids_kegg_map_associations.txt",header=FALSE,stringsAsFactors=FALSE)
> head(go)
V1 V2 V3
1 au5.g10000_t1 520259 map03010
2 au5.g10001_t1 520260 map00480

> go.terms=data.frame(go$V1,go$V3)
> head(go.terms)
go.V1 go.V3
1 au5.g10000_t1 map03010
2 au5.g10001_t1 map00480
3 au5.g10002_t1 <NA>
4 au5.g10003_t1 <NA>
5 au5.g10004_t1 <NA>
6 au5.g10005_t1 <NA>
> is.data.frame(go.terms)
[1] TRUE

> pwf=nullp(gene.vector,bias.data=gene.length)
>
> stats=goseq(pwf,gene2cat=go.terms,method='Wallenius')#good for R 2.12
Using manually entered categories.
Error in summary(map)[, 1] : incorrect number of dimensions
In addition: Warning message:
In goseq(pwf, gene2cat = go.terms, method = "Wallenius") :
Gene column could not be identified in gene2cat conclusively, using the one headed go.V1
>


If I construct

rownames(pwf) <- names(gene.length) ##this is what fixes it

I get the same error:

> pwf=nullp(gene.vector,bias.data=gene.length)
> rownames(pwf) <- names(gene.length) ##this is what fixes it
> stats=goseq(pwf,gene2cat=go.terms,method='Wallenius')#good for R 2.12
Using manually entered categories.
Error in summary(map)[, 1] : incorrect number of dimensions
Calls: goseq -> reversemapping
In addition: Warning message:
In goseq(pwf, gene2cat = go.terms, method = "Wallenius") :
Gene column could not be identified in gene2cat conclusively, using the one headed go.V1
Execution halted


Any suggestions welcomed - I'm missing something simple I suspect.

Charles

[crh]

I found the problem - different formatting of IDs imported:

all: au5.g1.t1

go.terms:au5.g1_t1


dope!

Charles

[tedtoal]

Error in summary(map)[, 1] : incorrect number of dimensions

Examples in the goseq pdf file dated 1 April 2010 show goseq() being called with the first argument "genes" and second argument "pwf". This is wrong. The "genes" argument is not used with goseq(). Its first argument is "pwf".
See ?goseq

[melis]

Hi all,
I am also having the same problem. My ID format is the same but still I get the same error when I run this.
> stats=goseq(pwf,gene2cat=go.terms,method='Wallenius')
Using manually entered categories.
Error in summary(map)[, 1] : incorrect number of dimensions
In addition: Warning message:
In goseq(pwf, gene2cat = go.terms, method = "Wallenius") :
Gene column could not be identified in gene2cat conclusively, using the one headed go.ID

Does anyone know the source of the error? What is "dimensions" mentioned in the message?

Thanks!
Melis

[upendra_35]

Examples in the goseq pdf file dated 1 April 2010 show goseq() being called with the first argument "genes" and second argument "pwf". This is wrong. The "genes" argument is not used with goseq(). Its first argument is "pwf".
See ?goseq

Thanks for the hint. I got into the same problem as everyone above and your hint helped me to fix it. Thanks