conver.snp.tped

[hakosalo]

I have a following problem. This might be because of I changed system from PC to MAC? Or maybe it is because of R version?
I have used files data1.tfam, data1.tped which were generated in a UNIX server
In my old PC/R 2.13 I gave the command

Code: Select all

convert.snp.tped(tped="sbt.tped", tfam="data1.tfam", out="data1.raw", strand="+")


and everything went smooth, but now, with same files made by our UNIX system (and with MAC / newest R 2.14)
the same command gives an error message

Code: Select all

Reading individual ids from file 'data1.tfam' ...
... done.  Read 113 individual ids from file 'data1.tfam'
Reading genotypes from file 'data1.tped' ...
Error in convert.snp.tped(tped = "data1.tped", tfam = "data1.tfam", out = "data1.raw",  :
  coding '0ֿ_?' for SNP not recognized !


Does anyone know why?

[lckarssen]

This might indeed have something to do with your switch from Unix to Mac, although I'm not sure. I know that Windows and Unix encode files in a different way (you can use the programs dos2unix and unix2dos on a Unix/Linux machine o convert between the two), maybe that goes for Mac as well. On the other hand, Mac OS X is based on a member of the Unix family...

Can you open a terminal on you Mac and try if the file command works? On my Linux system I use it to find out the encoding of a text file:

Code: Select all

$ file file_in_dos_format
file_in_dos_format: ASCII English text, with CRLF line terminators
$ file file_in_linux_format
file_in_linux_format: ASCII English text


Alternatively you can run less file.tped to see check by hand if the file is 'corrupt' and indeed has lines with '0ֿ_?'

[Hugoline]

Hi!

I got a similar error code when using the same command: "coding 'AG' for SNP not recognised!"

I checked the file in Linux, but can't figure out what's wrong. My tped.file looks "normal":
1 rs6675798 0 1166460 T T T T C C T T etc.

Missing values are depicted as 0.

Can anyone help me with this?

Thanks,

Hugoline

[Nicola Pirastu]

Have you checked that all SNPs name are like rsXXXXXXX?

Other people reporte this error and it turned that that was the problem.

I actually have no clue why this happens but it does.

Best.

Nicola

[Hugoline]

Hmmm... that's not it. I've constructed a dummy data-set and still got the "AG coding not recognized" error code.
The tped file runs in Plink, can there be any format error?

Best,

Hugoline

[marcin]

Hi,

UPDATE:
It actually turned out that the reading error was caused by specifying path in a relative way. Weird anyway...
using in convert.snp.ped
~kiero/Research/AtopyGSD/data/atopy.ped ---- more often works
~/Research/AtopyGSD/data/atopy.ped ---- sometimes works
/Users/kiero/Research/AtopyGSD/data/atopy.ped ---- so far always works...
When using tab-autocompletion, all the paths seem correct !!!

MESSAGE:
I previously reported a similar problem. I tried the suggestion from Nicola, but it worked only partially, i.e. error was reported on another error. What I did was t write a php script which read, trimmed and re-wrote every line. This helped with one dataset. I also tried writing it in Obj-C and Java and it ... didn't help. That is very weird. In addition, now sometimes I see a message "could not open file ..." (pertains to map file). This sometimes disappears after restarting R... Extra weird, esp. that I have really checked rights, umasks, and what not. I suspect it may be actually a bug in the newest R itself (file reading mechanism), not in GenABEL. As a role I observe such behavior after using awk on my map file. I should also say that I have OsX and ssd drive (errors coincide with the new R and the new disk, so maybe OS can't handle it [the disk] properly?). Will update you if I come up with stg. clever. For now try maybe to transpose your tied back to ped...

Marcin

[yurii]

RE: Hugoline

and still got the "AG coding not recognized" error code

can you please attach a toy TPED and TFAM files and exact sequence of commands so we could reproduce and look into the error?

Yurii

[Hugoline]

Hi,

Specifying the path differently does not seem to make a difference.


My commands are as following:

Code: Select all

convert.snp.tped(tped="test_2snp_5ID.tped", tfam="test_2snp_5ID.tfam",out="test_2snp_5ID.raw",bcast=1)

Code: Select all

 Reading individual ids from file 'test_2snp_5ID.tfam' ...
... done.  Read 5 individual ids from file 'test_2snp_5ID.tfam'
Reading genotypes from file 'test_2snp_5ID.tped' ...
Error in convert.snp.tped(tped = "test_2snp_5ID.tped", tfam = "test_2snp_5ID.tfam",  :
        coding 'AG' for SNP not recognised !

While trying to upload the dummy tped and tfam files, I got an error that "The extension tfam is not allowed". So I've copy-pasted the test data-set.

test_2snp_5ID.tfam:

Code: Select all

1   001   0   0   1   2
1   002   0   0   1   2
1   003   0   0   2   2
1   004   0   0   1   2
1   005   0   0   2   2

test_2snp_5ID.tped:

Code: Select all

1   rs3094315   0   742429   G G   A A   A A   G G   A A
1   rs3934834   0   995669   C C   C C   C C   T C   C C

Hugoline

[yurii]

This is strange; see what I get:

Code: Select all

> library(GenABEL)
> convert.snp.tped(tped="test_2snp_5ID.tped", tfam="test_2snp_5ID.tfam",out="test_2snp_5ID.raw",bcast=1)
Reading individual ids from file 'test_2snp_5ID.tfam' ...
... done.  Read 5 individual ids from file 'test_2snp_5ID.tfam'
Reading genotypes from file 'test_2snp_5ID.tped' ...
  ... read 1 lines ...
  ... read 2 lines ...
...done.  Read 2 SNPs from file 'test_2snp_5ID.tped'
Writing to file 'test_2snp_5ID.raw' ...
... done.
>


So it all work well. Zip-file with files you can run, and also outputs, attached (just try R -f runMe.R in the unpacked dir).

[rdlady]

I had a similar problem, when I tried to import my tped using the convert.snp.tped function from GenABEL I had the following error:
coding '' for SNP not recognised !

Or sometimes it would say:
coding '??' for SNP not recognised !

Or sometimes it would put a weir character between those quotes.

After a day of struggle I discovered that this problem was happening because the tped import function doesn't accept markers that have 100% missing data. I had in my tped files some markers that were like this: "0 rs123456 0 0 0 0 0 0 0 0 0 0 0 0 ... (all zeroes)".
So I used a simple reverse grep command in the bash terminal to remove all the markers that consisted only of zeroes. and now my convert.snp.tped works like a charm!

I hope this helps anybody that is having the same problem that I had.