R help - May 2008 - Unexpected behaviour in reading genomic coordinate files of R-2.7.0

Great R people,

I have noticed a strange behaviour in read.delim() and friends in the R 
2.7.0 version. I will describe you the problem and also the solution I 
already found, just to be sure it is an expected behaviour and also to 
tell people, who may experience the same difficulty, a way to overcome it.
And also to see if it is a proper behaviour or maybe a correction is needed.

Here is the problem:
I have some genomic coordinates files (bed files, a standard format, for 
example) containing a column (Strand) in which there is either a "+"
or
a "-".
In R-2.6.2patched (and every past version I have used) I never had 
problems in reading them in, as for example:
 > a <- read.table("coords.bed", skip=1)
 > disp(a)
class  data.frame
dimensions are  38650 6
first rows:
    V1        V2        V3    V4 V5 V6
1 chr1 100088396 100088446  seq1  0  +
2 chr1 100088764 100088814  seq2  0  -

If I do exactly the same command on the same file in R-2.7.0 the result 
I obtain is:
 > a <- read.table("coords.bed", skip=1)
 > disp(a)
class  data.frame
dimensions are  38650 6
first rows:
    V1        V2        V3    V4 V5 V6
1 chr1 100088396 100088446  seq1  0  0
2 chr1 100088764 100088814  seq2  0  0

and I completely loose the strand information, they are all zeros! I 
have also tried to put quotes around "+" and "-" in the file
before
reading it, to set in read.table() call stringsAsFactors=FALSE, to set 
"encoding" to a few different alternatives, but the result was always 
the same: they are all transformed in 0.

Then I tried scan() and I saw it was reading the character "+"
properly:
 > scan("coords.bed",  skip=1, nlines=1, what="ch")
Read 6 items
[1] "chr1"         "100088396"    "100088446.00"
"seq1" "0"
[6] "+"  

...my conclusion is that the lone "+" or "-" are not taken
as
"characters" in the data frame creation step, they are taken as 
"numeric" but, being without numbers are all converted to 0.
Is it correct if this behaviour happens also if they are surrounded by 
quotes?

Anyway, my temporary solution (which works without the need of changing 
the files) is:
a <- read.table("coords.bed", skip=1,
colClasses=c("character",
"numeric", "numeric", "character",
"numeric", "character"))
 > a[1:2,]
    V1        V2        V3    V4 V5 V6
1 chr1 100088396 100088446 seq1  0  +
2 chr1 100088764 100088814 seq2  0  -

Another way to avoid loosing strand information was to manually 
substitute an "R" to "-" and an "F" to
"+" in the file before reading it
in R. But it is much more cumbersome since the use of + and - is, for 
example, a standard format in bed files accepted and generated by the 
Genome Browser and other genome sites.

Please let me know what do you think. Ps. I saw this first in the Fedora 
version (rpm automatically updated), but it is reproduced also in the 
Windows version.

Thank you all people for your work and for making R the wonderful tool 
it is!

Cheers,

Margherita

-- 
 --
 -----------------------------------------------------------------------------------
 Margherita Mutarelli, PhD 
Seconda Universita' di Napoli
Dipartimento di Patologia Generale
 via L. De Crecchio, 7
 80138 Napoli - Italy
 Tel/Fax. +39.081.5665802

>From the NEWS file for R-patched:
     o   A field containing just a sign is no longer regarded as numeric
         (it was on all platforms in 2.7.0, but not on most in earlier
         versions of R).

So the default behaviour has already been changed.  The right way to 
overcome this was (as you discovered) to say what type you want the 
variables to be rather than let R guess for you.  But you cna also update 
your R (as the posting guide suggests).


On Wed, 28 May 2008, Margherita wrote:
> Great R people,
>
> I have noticed a strange behaviour in read.delim() and friends in the R
2.7.0
> version. I will describe you the problem and also the solution I already 
> found, just to be sure it is an expected behaviour and also to tell people,
> who may experience the same difficulty, a way to overcome it.
> And also to see if it is a proper behaviour or maybe a correction is
needed.
>
> Here is the problem:
> I have some genomic coordinates files (bed files, a standard format, for 
> example) containing a column (Strand) in which there is either a
"+" or a
> "-".
> In R-2.6.2patched (and every past version I have used) I never had problems
> in reading them in, as for example:
>> a <- read.table("coords.bed", skip=1)
>> disp(a)
> class  data.frame
> dimensions are  38650 6
> first rows:
>   V1        V2        V3    V4 V5 V6
> 1 chr1 100088396 100088446  seq1  0  +
> 2 chr1 100088764 100088814  seq2  0  -
>
> If I do exactly the same command on the same file in R-2.7.0 the result I 
> obtain is:
>> a <- read.table("coords.bed", skip=1)
>> disp(a)
> class  data.frame
> dimensions are  38650 6
> first rows:
>   V1        V2        V3    V4 V5 V6
> 1 chr1 100088396 100088446  seq1  0  0
> 2 chr1 100088764 100088814  seq2  0  0
>
> and I completely loose the strand information, they are all zeros! I have 
> also tried to put quotes around "+" and "-" in the file
before reading it, to
> set in read.table() call stringsAsFactors=FALSE, to set
"encoding" to a few
> different alternatives, but the result was always the same: they are all 
> transformed in 0.
>
> Then I tried scan() and I saw it was reading the character "+"
properly:
>> scan("coords.bed",  skip=1, nlines=1, what="ch")
> Read 6 items
> [1] "chr1"         "100088396"   
"100088446.00" "seq1" "0"          [6]
"+"
> ...my conclusion is that the lone "+" or "-" are not
taken as "characters" in
> the data frame creation step, they are taken as "numeric" but,
being without
> numbers are all converted to 0.
> Is it correct if this behaviour happens also if they are surrounded by 
> quotes?
>
> Anyway, my temporary solution (which works without the need of changing the
> files) is:
> a <- read.table("coords.bed", skip=1,
colClasses=c("character", "numeric",
> "numeric", "character", "numeric",
"character"))
>> a[1:2,]
>   V1        V2        V3    V4 V5 V6
> 1 chr1 100088396 100088446 seq1  0  +
> 2 chr1 100088764 100088814 seq2  0  -
>
> Another way to avoid loosing strand information was to manually substitute
an
> "R" to "-" and an "F" to "+" in the
file before reading it in R. But it is
> much more cumbersome since the use of + and - is, for example, a standard 
> format in bed files accepted and generated by the Genome Browser and other 
> genome sites.
>
> Please let me know what do you think. Ps. I saw this first in the Fedora 
> version (rpm automatically updated), but it is reproduced also in the
Windows
> version.
>
> Thank you all people for your work and for making R the wonderful tool it
is!
>
> Cheers,
>
> Margherita
>
> -- 
> --
> 
>
-----------------------------------------------------------------------------------
> Margherita Mutarelli, PhD Seconda Universita' di Napoli
> Dipartimento di Patologia Generale
> via L. De Crecchio, 7
> 80138 Napoli - Italy
> Tel/Fax. +39.081.5665802
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>
-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

This is not really addressing your problem, but I thought you might want to
know that the rtracklayer package in Bioconductor already supports parsing
BED files, as well as GFF and WIG. It's main purpose is to load the tracks
into genome browsers, like UCSC.

Michael

On Wed, May 28, 2008 at 1:11 AM, Margherita <atreeneedsaforest@yahoo.it>
wrote:
> Great R people,
>
> I have noticed a strange behaviour in read.delim() and friends in the R
> 2.7.0 version. I will describe you the problem and also the solution I
> already found, just to be sure it is an expected behaviour and also to tell
> people, who may experience the same difficulty, a way to overcome it.
> And also to see if it is a proper behaviour or maybe a correction is
> needed.
>
> Here is the problem:
> I have some genomic coordinates files (bed files, a standard format, for
> example) containing a column (Strand) in which there is either a
"+" or a
> "-".
> In R-2.6.2patched (and every past version I have used) I never had problems
> in reading them in, as for example:
> > a <- read.table("coords.bed", skip=1)
> > disp(a)
> class  data.frame
> dimensions are  38650 6
> first rows:
>   V1        V2        V3    V4 V5 V6
> 1 chr1 100088396 100088446  seq1  0  +
> 2 chr1 100088764 100088814  seq2  0  -
>
> If I do exactly the same command on the same file in R-2.7.0 the result I
> obtain is:
> > a <- read.table("coords.bed", skip=1)
> > disp(a)
> class  data.frame
> dimensions are  38650 6
> first rows:
>   V1        V2        V3    V4 V5 V6
> 1 chr1 100088396 100088446  seq1  0  0
> 2 chr1 100088764 100088814  seq2  0  0
>
> and I completely loose the strand information, they are all zeros! I have
> also tried to put quotes around "+" and "-" in the file
before reading it,
> to set in read.table() call stringsAsFactors=FALSE, to set
"encoding" to a
> few different alternatives, but the result was always the same: they are
all
> transformed in 0.
>
> Then I tried scan() and I saw it was reading the character "+"
properly:
> > scan("coords.bed",  skip=1, nlines=1, what="ch")
> Read 6 items
> [1] "chr1"         "100088396"   
"100088446.00" "seq1" "0"          [6]
> "+"
> ...my conclusion is that the lone "+" or "-" are not
taken as "characters"
> in the data frame creation step, they are taken as "numeric" but,
being
> without numbers are all converted to 0.
> Is it correct if this behaviour happens also if they are surrounded by
> quotes?
>
> Anyway, my temporary solution (which works without the need of changing the
> files) is:
> a <- read.table("coords.bed", skip=1,
colClasses=c("character", "numeric",
> "numeric", "character", "numeric",
"character"))
> > a[1:2,]
>   V1        V2        V3    V4 V5 V6
> 1 chr1 100088396 100088446 seq1  0  +
> 2 chr1 100088764 100088814 seq2  0  -
>
> Another way to avoid loosing strand information was to manually substitute
> an "R" to "-" and an "F" to "+" in
the file before reading it in R. But it
> is much more cumbersome since the use of + and - is, for example, a
standard
> format in bed files accepted and generated by the Genome Browser and other
> genome sites.
>
> Please let me know what do you think. Ps. I saw this first in the Fedora
> version (rpm automatically updated), but it is reproduced also in the
> Windows version.
>
> Thank you all people for your work and for making R the wonderful tool it
> is!
>
> Cheers,
>
> Margherita
>
> --
> --
>
>
-----------------------------------------------------------------------------------
> Margherita Mutarelli, PhD Seconda Universita' di Napoli
> Dipartimento di Patologia Generale
> via L. De Crecchio, 7
> 80138 Napoli - Italy
> Tel/Fax. +39.081.5665802
>
> ______________________________________________
> R-help@r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
> http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>
	[[alternative HTML version deleted]]