[R] filter a tab delimited text file

[Gabor Grothendieck]

On Fri, Sep 10, 2010 at 4:20 PM, Duke <duke.lists at gmx.com> wrote:
>  On 9/10/10 2:49 PM, Gabor Grothendieck wrote:
>>
>> On Fri, Sep 10, 2010 at 1:24 PM, Duke<duke.lists at gmx.com>  wrote:
>>>
>>>  Hi all,
>>>
>>> I have to filter a tab-delimited text file like below:
>>>
>>> "GeneNames"    "value1"    "value2"    "log2(Fold_change)"
>>>  "log2(Fold_change) normalized"    "Signature(abs(log2(Fold_change)
>>> normalized)>  4)"
>>> ENSG00000209350    4    35    -3.81131293562629    -4.14357714689656
>>>  TRUE
>>> ENSG00000177133    142    2    5.46771720082336    5.13545298955309
>>>  FALSE
>>> ENSG00000116285    115    1669    -4.54130810709955    -4.87357231836982
>>>  TRUE
>>> ENSG00000009724    10    162    -4.69995182667858    -5.03221603794886
>>>  FALSE
>>> ENSG00000162460    3    31    -4.05126372834704    -4.38352793961731
>>>  TRUE
>>>
>>> based on the last column (TRUE), and then write to a new text file,
>>> meaning
>>> I should get something like below:
>>>
>>> "GeneNames"    "value1"    "value2"    "log2(Fold_change)"
>>>  "log2(Fold_change) normalized"    "Signature(abs(log2(Fold_change)
>>> normalized)>  4)"
>>> ENSG00000209350    4    35    -3.81131293562629    -4.14357714689656
>>>  TRUE
>>> ENSG00000116285    115    1669    -4.54130810709955    -4.87357231836982
>>>  TRUE
>>> ENSG00000162460    3    31    -4.05126372834704    -4.38352793961731
>>>  TRUE
>>>
>>> I used read.table and write.table but I am still not very satisfied with
>>> the
>>> results. Here is what I did:
>>>
>>> expFC<- read.table( "test.txt", header=T, sep="\t" )
>>> expFC.TRUE<- expFC[expFC[dim(expFC)[2]]=="TRUE",]
>>> write.table (expFC.TRUE, file="test_TRUE.txt", row.names=FALSE, sep="\t"
>>> )
>>>
>>> Result:
>>>
>>> "GeneNames"    "value1"    "value2"    "log2.Fold_change."
>>>  "log2.Fold_change..normalized"
>>>  "Signature.abs.log2.Fold_change..normalized....4."
>>> "ENSG00000209350"    4    35    -3.81131293562629    -4.14357714689656
>>>  TRUE
>>> "ENSG00000116285"    115    1669    -4.54130810709955
>>>  -4.87357231836982
>>>  TRUE
>>> "ENSG00000162460"    3    31    -4.05126372834704    -4.38352793961731
>>>  TRUE
>>>
>>> As you can see, there are two points:
>>>
>>> 1. The headers were altered. All the special characters were converted to
>>> dot (.).
>>> 2. The gene names (first column) were quoted (which were not in the
>>> original
>>> file).
>>>
>> This will copy input lines matching pattern as well as the header to
>> the output verbatim preserving all quotes, spacing, etc.
>>
>> myFilter<- function(infile, outfile, pattern = "TRUE$") {
>>        L<- readLines(infile)
>>        cat(L[1], "\n", file = outfile)
>>        L2<- grep(pattern, L[-1], value = TRUE)
>>        for(el in L2) cat(el, "\n", file = outfile, append = TRUE)
>> }
>>
>> # e.g.
>> myFilter("infile.txt", "outfile.txt")
>>
>
> I love this the best! Even it is not as simple as the bash one liner
> (system( "cat infile.txt | grep -v FALSE > outfile.txt", wait=TRUE )), but I
> am very happy to learn that R does have other similar functions as in bash.
> If there is a document or a list of all such functions, that would be
> excellent.
>
> Thanks Gabor,
>

Check out these help files:

help.search(keyword = "character", package = "base")

-- 
Statistics & Software Consulting
GKX Group, GKX Associates Inc.
tel: 1-877-GKX-GROUP
email: ggrothendieck at gmail.com