[R] Fwd: Re: transpose and split dataframe
mccorm@ck @end|ng |rom mo|b|o@mgh@h@rv@rd@edu
Thu May 2 01:04:08 CEST 2019
Thank you very much, David and Jim for your work and solutions.
I have been working through both of them to better learn R. They both
proceed through a similar logic except David's starts with a character
matrix and Jim's with a dataframe, and both end with equivalent
dataframes ( identical(tmmdf, TF2list2)) returns TRUE ). They have
both been very helpful. However, there is one attribute of my intended
final dataframe that is missing.
Looking at part of the final dataframe:
AT1G69490 AT1G29860 AT1G29860.1 AT4G18170 AT4G18170.1 AT5G46350
1 *AT4G31950* *AT4G31950* AT5G64905 *AT4G31950* AT5G64905 *AT4G31950*
2 AT5G24110 AT5G24110 AT1G21120 AT5G24110 AT1G14540 AT5G24110
3 AT1G26380 AT1G05675 AT1G07160 AT1G05675 AT1G21120 AT1G05675
Row 1 has *AT4G31950* in columns 1,2,4 and 6, but AT4G31950 in columns 3
and 5. What I was aiming at would be that each row would have a unique
entry so that AT4G31950 is row 1 columns 1,2,4 and 6, and NA is row 1
columns 3 and 5. AT4G31950 is row 2 columns 3 and 5 and NA is row 2
columns 1,2,4 and 6. So, it would look like this:
AT1G69490 AT1G29860 AT1G29860.1 AT4G18170 AT4G18170.1 AT5G46350
1 AT4G31950 AT4G31950 NA AT4G31950 NA
2 NA NA AT4G31950 NA
I have been trying to adjust the code to get my intended result
basically by trying to build a dataframe one column at a time from each
entry in the character matrix, but have not got anything near working yet.
On 4/30/2019 6:29 PM, David L Carlson wrote
> If you read the data frame with read.csv() or one of the other read() functions, use the asis=TRUE argument to prevent conversion to factors. If not do the conversion first:
> # Convert factors to characters
> DataMatrix <- sapply(TF2list, as.character)
> # Split the vector of hits
> DataList <- sapply(DataMatrix[, 2], strsplit, split=",")
> # Use the values in Regulator to name the parts of the list
> names(DataList) <- DataMatrix[,"Regulator"]
> # Now create a data frame
> # How long is the longest list of hits?
> mx <- max(sapply(DataList, length))
> # Now add NAs to vectors shorter than mx
> DataList2 <- lapply(DataList, function(x) c(x, rep(NA, mx-length(x))))
> # Finally convert back to a data frame
> TF2list2 <- do.call(data.frame, DataList2)
> Try this on a portion of the list, say 25 lines and print each object to see what is happening.
> David L Carlson
> Department of Anthropology
> Texas A&M University
> College Station, TX 77843-4352
> -----Original Message-----
> From: R-help <r-help-bounces using r-project.org> On Behalf Of Matthew
> Sent: Tuesday, April 30, 2019 4:31 PM
> To: r-help using r-project.org
> Subject: [R] Fwd: Re: transpose and split dataframe
> Thanks for your reply. I was trying to simplify it a little, but must
> have got it wrong. Here is the real dataframe, TF2list:
> 'data.frame': 152 obs. of 2 variables:
> $ Regulator: Factor w/ 87 levels "AT1G02065","AT1G13960",..: 17 6 6 54
> 54 82 82 82 82 82 ...
> $ hits : Factor w/ 97 levels
> __truncated__,..: 65 57 90 57 87 57 56 91 31 17 ...
> And the first few lines resulting from dput(head(TF2list)):
> structure(list(Regulator = structure(c(17L, 6L, 6L, 54L, 54L,
> 82L), .Label = c("AT1G02065", "AT1G13960", "AT1G18860", "AT1G23380",
> "AT1G29280", "AT1G29860", "AT1G30650", "AT1G55600", "AT1G62300",
> "AT1G62990", "AT1G64000", "AT1G66550", "AT1G66560", "AT1G66600",
> "AT1G68150", "AT1G69310", "AT1G69490", "AT1G69810", "AT1G70510", ...
> This is another way of looking at the first 4 entries (Regulator is
> tab-separated from hits):
> So, the goal would be to
> first: Transpose the existing dataframe so that the factor Regulator
> becomes a column name (column 1 name = AT1G69490, column2 name
> AT1G29860, etc.) and the hits associated with each Regulator become
> rows. Hits is a comma separated 'list' ( I do not not know if
> technically it is an R list.), so it would have to be comma
> 'unseparated' with each entry becoming a row (col 1 row 1 = AT4G31950,
> col 1 row 2 - AT5G24410, etc); like this :
> ... I did not include all the rows)
> I think it would be best to actually make the first entry a separate
> dataframe ( 1 column with name = AT1G69490 and number of rows depending
> on the number of hits), then make the second column (column name =
> AT1G29860, and number of rows depending on the number of hits) into a
> new dataframe and do a full join of of the two dataframes; continue by
> making the third column (column name = AT1G2986) into a dataframe and
> full join it with the previous; continue for the 152 observations so
> that then end result is a dataframe with 152 columns and number of rows
> depending on the entry with the greatest number of hits. The full joins
> I can do with dplyr, but getting up to that point seems rather difficult.
> This would get me what my ultimate goal would be; each Regulator is a
> column name (152 columns) and a given row has either NA or the same hit.
> This seems very difficult to me, but I appreciate any attempt.
> On 4/30/2019 4:34 PM, David L Carlson wrote:
>> External Email - Use Caution
>> I think we need more information. Can you give us the structure of the data with str(YourDataFrame). Alternatively you could copy a small piece into your email message by copying and pasting the results of the following code:
>> The data frame you present could not be a data frame since you say "hits" is a factor with a variable number of elements. If each value of "hits" was a single character string, it would only have 2 factor levels not 6 and your efforts to parse the string would make more sense. Transposing to a data frame would only be possible if each column was padded with NAs to make them equal in length. Since your example tries use the name TF2list, it is possible that you do not have a data frame but a list and you have no factor levels, just character vectors.
>> If you are not familiar with R, it may be helpful to tell us what your overall goal is rather than an intermediate step. Very likely R can easily handle what you want by doing things a different way.
>> David L Carlson
>> Department of Anthropology
>> Texas A&M University
>> College Station, TX 77843-4352
>> -----Original Message-----
>> From: R-help<r-help-bounces using r-project.org> On Behalf Of Matthew
>> Sent: Tuesday, April 30, 2019 2:25 PM
>> To: r-help (r-help using r-project.org)<r-help using r-project.org>
>> Subject: [R] transpose and split dataframe
>> I have a data frame that is a lot bigger but for simplicity sake we can
>> say it looks like this:
>> Regulator hits
>> AT1G69490 AT4G31950,AT5G24110,AT1G26380,AT1G05675
>> AT2G55980 AT2G85403,AT4G89223
>> In other words:
>> data.frame : 2 obs. of 2 variables
>> $Regulator: Factor w/ 2 levels
>> $hits : Factor w/ 6 levels
>> I want to transpose it so that Regulator is now the column headings
>> and each of the AGI numbers now separated by commas is a row. So,
>> AT1G69490 is now the header of the first column and AT4G31950 is row 1
>> of column 1, AT5G24110 is row 2 of column 1, etc. AT2G55980 is header of
>> column 2 and AT2G85403 is row 1 of column 2, etc.
>> I have tried playing around with strsplit(TF2list[2:2]) and
>> strsplit(as.character(TF2list[2:2]), but I am getting nowhere.
>> R-help using r-project.org mailing list -- To UNSUBSCRIBE and more, see
>> PLEASE do read the posting guidehttp://www.R-project.org/posting-guide.html
>> and provide commented, minimal, self-contained, reproducible code.
> [[alternative HTML version deleted]]
> R-help using r-project.org mailing list -- To UNSUBSCRIBE and more, see
> 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]]
More information about the R-help