Re: [Chipster-tech] General bug when selecting column of phenodata (chipster 3.?)

[<chi...@cs...>]

Hi Oliver,

Thank you for reporting this and sorry for the delay due to holidays. We will look into this soon.

Best,
Eija

----- Original Message -----
From: "Oliver Heil" <o....@dk...>
To: "chipster-tech" <chi...@li...>
Sent: Wednesday, 11 August, 2021 12:56:08
Subject: [Chipster-tech] General bug when selecting column of phenodata (chipster 3.?)

Dear chipster developers,

There is a general bug in the (microarray) tools, when there is a
parameter for the tool to specify a column of the phenodata.

(currently I am still on chipster 3)

E.g. in stat-two-groups.R

> phenodata<-read.table("phenodata.tsv", header=T, sep="\t")
> groups<-phenodata[,pmatch(column,colnames(phenodata))]

When the user defines a new column which contains "-" (minus sign),
e.g. "group-1", read.table transforms this column name into "group.1"
for the resulting data.frame. With that pmatch will return NA, because
the column parameter will not match the colnames of phenodata.

Above two lines are the generic code in several tools.

Instead of

> phenodata<-read.table("phenodata.tsv", header=T, sep="\t")
I suggest the following generic code:

> phenoheader=strsplit(readLines("phenodata.tsv",n=1),split="\t")
> phenodata=read.table("phenodata.tsv", header=T, sep="\t")
> colnames(phenodata)=phenoheader[[1]]
at least, when a match on the column names is done afterwards.

Best regards,

Oli




_______________________________________________
Chipster-tech mailing list
Chi...@li...
https://lists.sourceforge.net/lists/listinfo/chipster-tech