How to plot dendrogram based on sample names

Question

[Bioinfonext]

Hi,

I do have FPKM count and interested in dendrogram for samples cluster.

I used below code but it generate dendogram based on geneID instead of sampleID.

  > countMatrix = read.table("Trinity_trans.counts.matrix.txt",header=T,sep='\t',check.names=F,row.names=1)

> dim(countMatrix)

[1] 142686      6

> head(countMatrix)

                          AS_0DAP AS_4DAP AS_8DAP NMK_0DAP NMK_4DAP NMK_8DAP

TRINITY_DN17944_c0_g1_i11   14.32   24.63    8.21     4.54       20     8.49

TRINITY_DN7591_c0_g1_i1      0.00    0.00    1.00     3.00        3     0.00

TRINITY_DN28918_c0_g1_i1     1.00    2.00    1.00     0.00        2     0.00

TRINITY_DN14082_c2_g2_i5     6.00    5.00    1.00     0.00        1     0.00

TRINITY_DN31994_c0_g1_i1     1.00    2.00    0.00     0.00        0     3.00

TRINITY_DN19560_c0_g1_i1     1.00    3.00    0.00     0.00        1     1.00

> rv <- rowVars(countMatrix)

> summary(rv)

     Min.   1st Qu.    Median      Mean   3rd Qu.      Max. 

0.000e+00 1.000e+00 1.500e+01 3.570e+05 5.180e+02 4.122e+09 


> (q75 <-quantile(rowVars(countMatrix), .75))
     75% 

518.3202 

> m2 <- countMatrix[rv >q75, ]

> dim(m2)

[1] 35672     6

> summary(rowVars(m2))

     Min.   1st Qu.    Median      Mean   3rd Qu.      Max. 

5.180e+02 1.670e+03 6.677e+03 1.428e+06 4.101e+04 4.122e+09 

> d <- dist(m2, method="euclidean")

> h <-hclust(d, method="complete")

> plot(h)

I will be thankful for your time and help.

Regards

nabiyogesh

Answer 1

[Kevin Blighe]

Hey, you may simply have to transpose your data at some point. This can be done via the t() function in R.

[Bioinfonext]

thank a lot, could you please also suggest me what this value means here: 518.3202

 (q75 <-quantile(rowVars(countMatrix), .75))
     75% 

518.3202

[Kevin Blighe]

May I ask from where you found that line of code?

[Bioinfonext]

I used genefilter to use only most variable:

https://support.bioconductor.org/p/61678/

[Kevin Blighe]

Oh, I see, it is just filtering out genes based on a cut-off of variance. Genes of low variance add 'no' information to the differential expression test, so, many filter these out. To put it another way: with that code, you are only retaining the genes that are in the upper quartile of the variance range.

[Bioinfonext]

Thanks for your all help and time.