Juanhijuan
Juanhijuan

Reputation: 102

Anova test in the loop

I had table like this:

    id BiotinControl1_2 BiotinControl2 BiotinControl3 BiotinTreatment1_2 BiotinTreatment2 BiotinTreatment3 Sequence
1   75       3893050.50     2717893.32      3206861.1         3435216.40        3768203.0 3647604.45 AAAAGAAAVANQGKK
2  192        900604.61      741299.33       937413.2          818936.89         937764.7 751303.46  AAAAGAAAVANQGKK
3 3770         90008.14       87127.07       107568.6           85120.95         101947.6 90152.82   AAFTKLDQVWGSE

I used the code below to reorganize my data:

tbl_reo <- melt(tbl_anv, measure.vars=2:7)

So now my data looks like below:

    id          Sequence         variable      value
1   75   AAAAGAAAVANQGKK BiotinControl1_2 3893050.50 
2  192   AAAAGAAAVANQGKK BiotinControl1_2  900604.61 
3 3770     AAFTKLDQVWGSE BiotinControl1_2   90008.14

I want to get an anova analysis. I want to do it id by id in a loop. So firstly I created a table for the id containing all 6 values and the variable column which states to which variable the value belongs to. I want to do a anova calculation for those 2 columns.

Edit: So for each id , I'd like to compute the lm(value~variable).

aov.test <- summary(aov(tbl_reo$value ~ as.factor(tbl_reo$variable)))

tbl_reo[,5]  <- aov.test[[1]]$'Pr(>F)'[1]

Code which I used to calculate the anova but it's not working properly. I'd like to put the results in an extra column in my data.

tbl_anv <- tbl_all_onlyK[,c("id", "BiotinControl1_2", "BiotinControl2", "BiotinControl3",    "BiotinTreatment1_2", "BiotinTreatment2", "BiotinTreatment3", "Sequence")]

tbl_reo <- melt(tbl_anv, measure.vars=2:7)

dat <- vector("integer", length = ncol(tbl_reo))
names(dat) <- colnames(tbl_reo)

for (var in variable) {
dat[var] <- anova(lm(value ~ tbl_reo[, var], data = tbl_reo))$"Pr(>F)"[1]
}

Upvotes: 0

Views: 2831

Answers (1)

agstudy
agstudy

Reputation: 121608

Very hard to understand what do you want to do. Since you question is still not clear this is just an approximated answer....

But I think you are looking for by or for ddply from plyr package, to do some treatment by group. For example, First I create some reproducible data. (Please learn how to reproduce data, it is essential to resolve your problem).

set.seed(1)
vars <- c("id", "BiotinControl1_2", "BiotinControl2", "BiotinControl3",   
         "BiotinTreatment1_2", "BiotinTreatment2", "BiotinTreatment3",
         "Sequence")
tbl_reo <- data.frame(value   = rnorm(100),
                      id = gl(6,100/6),
                      variable= sample(vars,100,rep=TRUE))

So my data is lokking like this :

str(tbl_reo)
'data.frame':   100 obs. of  3 variables:
 $ value   : num  -0.626 0.184 -0.836 1.595 0.33 ...
 $ id      : Factor w/ 6 levels "1","2","3","4",..: 1 1 1 1 1 1 1 1 1 1 ...
 $ variable: Factor w/ 8 levels "BiotinControl1_2",..: 2 1 4 2 1 4 4 1 2 5 ...

So for each id , I'd like to compute the lm(value~variable) and compare anova between ids...

So using by you can get the anova for each id like this:

by(tbl_reo,tbl_reo$id,function(x){
  anova(lm(value ~ variable, data = x))$"Pr(>F)"[1]
})

tbl_reo$id: 1
[1] 0.7430758
--------------------------------------------------------------------------------------------------------- 
tbl_reo$id: 2
[1] 0.122237
--------------------------------------------------------------------------------------------------------- 
tbl_reo$id: 3
[1] 0.8914668
--------------------------------------------------------------------------------------------------------- 
tbl_reo$id: 4
[1] 0.7790441
--------------------------------------------------------------------------------------------------------- 
tbl_reo$id: 5
[1] 0.6833726
--------------------------------------------------------------------------------------------------------- 
tbl_reo$id: 6
[1] 0.7323833

Upvotes: 3

Related Questions