R: Rearrange specific rows of single-column data frame into new column

Question

EDIT : Thought it might be useful to mention for anyone searching later that this data is the output I got from Biomart when I tried to output sequence attributes.

I have the following genomic data:

structure(list(Sequences = structure(c(2L, 10L, 3L, 8L, 9L, 1L, 
5L, 4L, 6L, 7L), .Label = c(">ENSRNOG00000000902|Hsph1", ">ENSRNOG00000001136|Pebp1", 
">ENSRNOG00000001214|Pfkl", "AGAGAGGCGAGCGGCGGAGAGCGGTGGCAAATACTGAACGCAGTCTCGCAGGGTAAGCCC", 
"GAGCGATTGGGACCTCCCCTTTTGGATTGGTAGCTGAGCGGCAGTGGCGGCGGCTGCGTG", 
"GAGGCATCTTCCCGGCCGGTCGGGAGCAGGAGGAGCACGCAGCGGATCCCAGGCAGAGGC", 
"GGACCGGGCCAGCC", "GGCGGGGACAGGCGACAGCCGCGCGGAACGCAGAGCGGCGGGAGAGGAGCTCGGGCTCCT", 
"GGTCTCTGCTGCCGTC", "GTTTAACTGCACTCGGGACTCGGCGCGCGCGTGTGTCTGTTCTCTCCATCGTC"
), class = "factor")), .Names = "Sequences", class = "data.frame", row.names = c(NA, 
-10L))

I would like to rearrange the data so the first column shows the gene ID info (e.g. for the first case it would be: ">ENSRNOG00000001136|Pebp1") and then the lines of genomic code underneath it would appear together in the column next to it. Note that rows 7-10 have multiple lines of genetic code. Here, all the strings below the gene ID info would be concatenated together into one line instead of spread over 4 separate rows. Lastly, I also want to remove the ">" symbol that comes before each of the gene IDs.

The final output would thus be:

ID                          Sequence
ENSRNOG00000001136|Pebp1    GTTTAACTGCACTCGGGACTCGGCGCGCGCGTGTGTCTGTTCTCTCCATCGTC
ENSRNOG00000001214|Pfkl GGCGGGGACAGGCGACAGCCGCGCGGAACGCAGAGCGGCGGGAGAGGAGCTCGGGCTCCTGGTCTCTGCTGCCGTC
ENSRNOG00000000902|Hsph1    GAGCGATTGGGACCTCCCCTTTTGGATTGGTAGCTGAGCGGCAGTGGCGGCGGCTGCGTGAGAGAGGCGAGCGGCGGAGAGCGGTGGCAAATACTGAACGCAGTCTCGCAGGGTAAGCCCGAGGCATCTTCCCGGCCGGTCGGGAGCAGGAGGAGCACGCAGCGGATCCCAGGCAGAGGCGGACCGGGCCAGCC

Note that this is just the first few lines of ~2500 row data frame. The solution needs to be general enough that it can parse situations where the number of rows taken up by the sequence may be more than the 4 rows shown in the example above.

Answer 1

After labeling rows with whether it's a gene name (based on whether it has ">"), you can use cumsum to determine groups of the rows. Then you can use substr to remove the leading ">" and tapply with paste to combine all the genes in each group.

dat$Sequences <- as.character(dat$Sequences)
(is.gene <- grepl(">", dat$Sequences))
# [1]  TRUE FALSE  TRUE FALSE FALSE  TRUE FALSE FALSE FALSE FALSE
(gene.groups <- cumsum(is.gene))
# [1] 1 1 2 2 2 3 3 3 3 3
data.frame(ID=substr(dat$Sequences[is.gene], 2, nchar(dat$Sequences[is.gene])),
           Sequence=tapply(dat$Sequences[!is.gene], gene.groups[!is.gene], paste, collapse=""))
#                         ID
# 1 ENSRNOG00000001136|Pebp1
# 2  ENSRNOG00000001214|Pfkl
# 3 ENSRNOG00000000902|Hsph1
#                                                                                                                                                                                             Sequence
# 1                                                                                                                                              GTTTAACTGCACTCGGGACTCGGCGCGCGCGTGTGTCTGTTCTCTCCATCGTC
# 2                                                                                                                       GGCGGGGACAGGCGACAGCCGCGCGGAACGCAGAGCGGCGGGAGAGGAGCTCGGGCTCCTGGTCTCTGCTGCCGTC
# 3 GAGCGATTGGGACCTCCCCTTTTGGATTGGTAGCTGAGCGGCAGTGGCGGCGGCTGCGTGAGAGAGGCGAGCGGCGGAGAGCGGTGGCAAATACTGAACGCAGTCTCGCAGGGTAAGCCCGAGGCATCTTCCCGGCCGGTCGGGAGCAGGAGGAGCACGCAGCGGATCCCAGGCAGAGGCGGACCGGGCCAGCC

Answer 2

Here's a possible data.table solution (though you will have additional column called indx that you can remove later if you wish)

library(data.table)
setDT(df)[, list(
              ID = sub("^>", "", Sequences[1L]), 
              Sequences = paste(Sequences[-1L], collapse = "")
              ), 
            by = list(indx = cumsum(grepl(">", Sequences)))]