Amova analysis

Commands in mothur
#1

Hei,

I run the Amova analysis on three groups, the overall P value was significant among groups, but no significant different for pairwised group. Any help for explanation?

152y-2y-60y Among Within Total
SS 0.540043 0.964886 1.50493
df 2 6 8
MS 0.270021 0.160814

Fs: 1.67909
p-value: 0.003*

152y-2y Among Within Total
SS 0.286785 0.626688 0.913472
df 1 4 5
MS 0.286785 0.156672

Fs: 1.83048
p-value: 0.087

152y-60y Among Within Total
SS 0.276427 0.652278 0.928705
df 1 4 5
MS 0.276427 0.163069

Fs: 1.69515
p-value: 0.098

2y-60y Among Within Total
SS 0.246852 0.650806 0.897659
df 1 4 5
MS 0.246852 0.162702

Fs: 1.51721
p-value: 0.094

Many thanks,

Hui

#2

You probably don’t have sufficient power to detect a pairwise difference. It looks like you may only have 2 or 3 samples per group.

Pat

#3

can you tell me what does it mean those shorts from exactly SS, df, and MS?

if I want to show the variance between the sample and within the sample in which command do I need to give instructions?

Look here is an output I have a total of 12 samples of 4 groups of biological replicates. Actually, I cannot explain my output there is an " Fs: -nan " why f-statistics is -nan?

A1-A2-A3-A4-B1-B2-B3-B4-C1-C2-C3-C4
SS	3.6387	0	3.6387
df	11	0	11
MS	0.330791	-nan

Fs:	-nan
p-value: 1

Experiment-wise error rate: 0.05
If you have borderline P-values, you should try increasing the number of iterations

Please tell me how can I adjust P-Value Amova output?

#4

Can you post your design file? It looks like you have 12 groups. You have to use the design file to group your samples. Look at the one we use in the MiSeq SOP for an example.

SS = sum of squared errors
df = degrees of freedom
MS = mean squared errors

Pat