I have data files structured like this:
OTU1 PIA0 1120
OTU2 PIA1 2
OTU2 PIA3 6
OTU2 PIA4 10
OTU2 PIA5 1078
OTU2 PIN1 24
OTU2 PIN2 45
OTU2 PIN3 261
OTU2 PIN4 102
OTU3 PIA0 16
OTU3 PIA1 59
OTU3 PIA2 27
OTU3 PIA3 180
OTU3 PIA4 200
OTU3 PIA5 251
OTU3 PIN0 36
OTU3 PIN1 61
OTU3 PIN2 156
OTU3 PIN3 590
OTU3 PIN4 277
OTU4 PIA0 401
OTU4 PIN0 2
And I want to create a matrix that shows combination of data from the second column taking the first column as reference for the counts of combination (showing how many times, measured each one by the first column number -OTU1, OTU2, OTU3, OTU4- each datum from the second column appears together with each other in the same OTU). It needs to look like this:
PIA0 PIA1 PIA2 PIA3 PIA4 PIA5 PIN0 PIN1 PIN2 PIN3 PIN4
PIA0 1 1 1 1 1 1 2 1 1 1 1
PIA1 1 0 1 2 2 2 1 2 2 2 2
PIA2 1 1 0 1 1 1 1 1 1 1 1
PIA3 1 2 1 0 2 2 1 2 2 2 2
PIA4 1 2 1 2 0 2 1 2 2 2 2
PIA5 1 2 1 2 2 0 1 2 2 2 2
PIN0 2 1 1 1 1 1 0 1 1 1 1
PIN1 1 2 1 2 2 2 1 0 2 2 2
PIN2 1 2 1 2 2 2 1 2 0 2 2
PIN3 1 2 1 2 2 2 1 2 2 0 2
PIN4 1 2 1 2 2 2 1 2 2 2 0
Data shared between a row and a column with the same name reflects the number of times this datum appears alone in an OTU.
Any ideas?
I have read about R libraries 'reshape2' and command 'acast' here, but with that I can only change the shape of a matrix with all data in it, not make combination counts as desired. I have also been thinking about a Biopython script, but I think it would be too big and difficult to write it down with my little knowledge about programming.
The goal is to build a matrix like the one in the example so I can run CIRCOS online program with these data.
You can use
dcast
to create a binary matrix indicating the presence of each PI inside each OTU, and then multiply it by itself to have the counts.