I have a dataset of binary data, and my PhD supervisor asked me to compute an AMOVA on it. I'm not familiar with such a test, furthermore the error messages are not convenient (private joke).
So If I understood I need a frequency of each marker in every group so I have 3 groups.
V1 V2 V3
1 0 0 5
2 33 35 32
3 25 25 24
4 13 20 33
5 67 65 61
6 0 0 1
7 79 85 80
…
Then I get a matrix of euclidean distance between sample:
B-40-2x-T B-41-2x-T B-43-2x-T B-47-2x-T B-52-2x-T B-72-3x-T B-73-2x-A
B-41-2x-T 3.000000
B-43-2x-T 2.828427 3.316625
B-47-2x-T 3.872983 4.242641 4.123106
B-52-2x-T 3.464102 3.872983 3.741657 3.000000
B-72-3x-T 3.741657 3.872983 3.741657 3.316625 2.828427
B-73-2x-A 3.162278 3.000000 3.162278 3.605551 2.828427 2.828427
B-75-4x-T 3.872983 4.472136 4.358899 4.000000 4.582576 4.582576
Well, I get the usual "Erreur dans amova(frec, distE) : Non convenient samples"
Looking at the example I did something obviously wrong, because the rownames of the "sample" should agree with the names of "distances" matrix…
So I am lost, I try to get some analyse of the groups of individual not groups of alleles…
And calculating the "distance" between alleles in function of the sample that present them doesn't seem to make sense…
So who could help me in understanding how this test should work? how to understand the basis. How to compute it and how to analyse output?
Thanks in advance.
Hi Poornima,
As far as i can see you have:
1. Missing values in your dataset (NAs), which you should delete (or complete) them before you run the distances
2. You do not provide a structures dataset, which should be a DataFrame
please check the following code
########## R - code #############
samp
@poornima: AMOVA is analyse on molecular variance
@Kostas:
There isn't any missing value, I just proposed a tiny part on a 200x200matrix to give an aspect of it ;)
I did not provide structure because from ?amova:
"samples
a data frame with haplotypes (or genotypes) as rows, populations as columns and abundance as entries"
Then I should give abondance (then summing/getting frequency according to populations) that mean I only have 3 populations left without any structure to gather them …
hi Benjamin,
I apologise for any unintentional mistake with your name, due to lack of time, in my previous note.
I am not sure I understand your reasoning for the structures dataframe but it does not really matter, as it should run without it.
The only way I can reproduce your error is when the dataset is not in a "data.frame" object, but in a matrix
So, please provide us with the output of the command
class(YourDataSet)
Well, I had this error :p but I quickly figured that out with the help of the exemple:
> class(frec)
[1] "data.frame"
But I sure it is the way of doing my object that isn't good… I start from an object name "Betula" looking like that:
al270 al273 al121 al122 al123 al132 al134 al137 al146 al163.1 al217 al229
B-40-2x-T 0 0 0 1 1 0 0 0 1 0 0 0
B-41-2x-T 1 0 0 1 1 0 0 0 1 0 1 0
B-43-2x-T 0 0 1 1 1 0 0 0 1 0 0 0
B-47-2x-T 1 0 0 1 0 0 1 0 0 0 0 0
B-52-2x-T 1 0 0 1 1 0 0 0 0 0 0 0
B-72-3x-T 1 0 0 1 0 0 0 0 0 0 0 0
B-73-2x-A 1 0 0 1 1 0 0 0 1 0 1 0
B-75-4x-T 0 0 1 1 0 0 0 0 1 0 0 0
B-83-4x-A 0 0 1 1 1 1 0 0 1 0 0 0
(of course it is just a tiny random part of it)
Then I tried to prepare my sample object…
#prep data
apply(Betula[code$Ploidy==2,],2,sum)->num
nrow(Betula[code$Ploidy==2,])->dem
round(100*num/dem)->P2
apply(Betula[code$Ploidy==3,],2,sum)->num
nrow(Betula[code$Ploidy==3,])->dem
round(100*num/dem)->P3
apply(Betula[code$Ploidy==4,],2,sum)->num
nrow(Betula[code$Ploidy==4,])->dem
round(100*num/dem)->P4
frec
OK. Lets do it step by step
1. you need a data.frame to be you "samples" object. You are using the data.frame named "frec" which has a (?) number of rows (without row.names) and 3 columns named "V1","V2", "V3", with numerical values.
2. Then you are using a second object named "gel" to produce the distances (distE).
Does the "gel" object have the same number of rows and columns with "frec" data.frame??? If not the error is "Error in amova(frec, distE) : Non convenient samples"
Also you standardize it for presence-absence "binary=TRUE". Does this give you a second error like: "In is.euclid(distances) : Zero distance(s)" ????
Well… The "gel" object is either betula or t(betula) in function of the computer out put. In the present case, they are just identical.
number of row of "frec" is equal to number of row of betula (or gel) id est the number of marker…
Then distE is the euclidian distance between sample, so number of row= number of column = number of sample
I do not have any "In is.euclid(distances) : Zero distance(s)" like error…
hi,
vegdist requires a community matrix and not a data.frame.
Therefore the number of rows will not be the right one.
check the results of the following code:
################ R-code ######################
gel
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Software
- R