[R] biclustering using grouped columns

Nacho Caballero nachocab at gmail.com
Thu Jan 21 09:08:23 CET 2010


Hi,
I'm trying to perform biclustering to a matrix with genes as rows and
columns representing the frequency of each codon for each of its
aminoacids, like:

--------  Tyr(TAT)  Tyr(TAC)       Iso(ATT) Iso(ATC) Iso(ATA)
Phe(TTT) Phe(TTC)
gene1      0.2           0.8               0.3          0.3
0.1               0.4           0.6
gene2      0.8           0.2               0.1          0.4
0.5               0.6           0.4
gene3      0.3           0.7               0.3          0.3
0.1               0.4           0.6

The problem is that these columns should be somehow joined, since a
bicluster that only contains Tyr(TAC) and Iso(ATC) is meaningless;
only one that would contain ALL the columns for each AA makes sense
(ex: all Tyr(1,2) and all Iso(3,4,5))

I'm trying to find a way to combine the columns of each AA into a
single number, with the additional requirement that the difference
between rows should remain constant
Ex: gene3 - gene1 for Tyr:
0.3 - 0.2 => 0.1
0.8 - 0.7 => 0.1
so I can't do something like join both numbers and then substract
0.2 | 0.8 => 28,
0.3 | 0.7 => 37,
37 - 28 => 9 WRONG)

Any ideas? I really appreciate your help,

Nacho



More information about the R-help mailing list