lofleoflx.md

title	author	output	subtitle	tags	abstract
R Notebook for effects of fluoxetine on zebrafish aggression (LaNeC)	Caio Maximino^[Universidade Federal do Sul e Sudeste do Pará] Monica Gomes Lima^[Universidade do Estado do Pará] Hellen Barbosa^[Universidade Federal do Sul e Sudeste do Pará]	github_document	From project "Role of the serotonergic system on aggressive behavior in two zebrafish phenotypes"	aggression zebrafish fluoxetine	Behavioral data in the mirror-induced aggression test using fluoxetine in two zebrafish phenotypes (longfin and leopard)

This is an R Markdown Notebook for the data analysis of the research project "Role of the serotonergic system on aggressive behavior in two zebrafish phenotypes".

Data is produced by members from Laboratório de Neurociências e Comportamento "Frederico Guilherme Graeff", affiliated to Universidade Federal do Sul e Sudeste do Pará and Universidade do Estado do Pará. The package will include primary data for a behavioral experiment on the effects of fluoxetine on zebrafish aggressive behavior. In Experiment 1, basline aggression levels were compared between phenotype; in Experiment 2, the effects of fluoxetine were tested.

When you execute code within the notebook, the results appear beneath the code.

• Load needed libraries:

if(!require(ggplot2)){
    install.packages("ggplot2")
    library(ggplot2)
}
if(!require(coin)){
    install.packages("coin")
    library(coin)
}
if(!require(RCurl)){
    install.packages("RCurl")
    library(RCurl)
}

if(!require(plyr)){
    install.packages("plyr")
    library(plyr)
}

if(!require(rcompanion)){
    install.packages("rcompanion")
    library(rcompanion)
}

if(!require(WRS2)){
    install.packages("WRS2")
    library(WRS2)
}

if(!require(psych)){
    install.packages("psychh")
    library(psych)
}

• Load data

x1 <- getURL("https://raw.githubusercontent.com/lanec-unifesspa/5HT-aggression/master/exp1.csv")
exp1 <- read.csv(text = x1)
exp1$Phenotype <- as.factor(exp1$Phenotype)
View(exp1)

x2 <- getURL("https://raw.githubusercontent.com/lanec-unifesspa/5HT-aggression/master/exp2.csv")
exp2 <- read.csv(text = x2)
exp2$Phenotype <- as.factor(exp2$Phenotype)
exp2$Dose <- as.factor(exp2$Dose)
View(exp2)

• Run Approximative Two-Sample Fisher-Pitman Permutation Tests on data from Experiment 1

1. Latency data

oneway_test(Latency ~ Phenotype, data = exp1, distribution="approximate"(B=10000))

2. Display duration data

oneway_test(Dur.Display ~ Phenotype, data = exp1, distribution="approximate"(B=10000))

3. Display frequency data

oneway_test(N.Display ~ Phenotype, data = exp1, distribution="approximate"(B=10000))

4. Time near mirror data

oneway_test(T.Mirror ~ Phenotype, data = exp1, distribution="approximate"(B=10000))

5. Locomotion data

oneway_test(N.Quad ~ Phenotype, data = exp1, distribution="approximate"(B=10000))

Produce figures on ggplot2

ggplot(exp1, aes(x = factor(Phenotype), y = Latency, colour = Phenotype)) + geom_boxplot() + geom_jitter(position = "dodge", alpha = 0.3) + labs(x = "", y = "Latency to first display (s)", color = "Phenotype")

ggplot(exp1, aes(x = factor(Phenotype), y = Dur.Display, colour = Phenotype)) + geom_boxplot() + geom_jitter(position = "dodge", alpha = 0.3) + labs(x = "", y = "Display duration (s)", color = "Phenotype")

ggplot(exp1, aes(x = factor(Phenotype), y = N.Display, colour = Phenotype)) + geom_boxplot() + geom_jitter(position = "dodge", alpha = 0.3) + labs(x = "", y = "Display frequency (N)", color = "Phenotype")

ggplot(exp1, aes(x = factor(Phenotype), y = T.Mirror, colour = Phenotype)) + geom_boxplot() + geom_jitter(position = "dodge", alpha = 0.3) + labs(x = "", y = "Time near mirror (s)", color = "Phenotype")

ggplot(exp1, aes(x = factor(Phenotype), y = N.Quad, colour = Phenotype)) + geom_boxplot() + geom_jitter(position = "dodge", alpha = 0.3) + labs(x = "", y = "Squares crossed (N)", color = "Phenotype")

• Run bootstrapped ANOVA for main and interaction effects on 2-way ANOVA, in Experiment 2 (based on https://rcompanion.org/rcompanion/d_08a.html)

1. Latency data 1.1) Run two-way ANOVAs on M-estimators

pbad2way(Latency ~ Phenotype + Dose + Phenotype:Dose, data = exp2, est="mom", nboot = 5000)

1.2) Produce post-hoc tests for main effects

postLat <- mcp2a(Latency ~ Phenotype + Dose + Phenotype:Dose, data = exp2, est = "mom", nboot = 5000)
postLat$contrasts
postLat

1.3) Produce post-hoc tests for interaction effect

exp2$Factor.int <- interaction(exp2$Dose, exp2$Phenotype)
exp2$Factor.int <- factor(exp2$Factor.int, levels = c("0 mg/kg.LOF", "2.5 mg/kg.LOF", "5.0 mg/kg.LOF", "0 mg/kg.LEO", "2.5 mg/kg.LEO", "5.0 mg/kg.LEO"))
headTail(exp2)
PTLat <- pairwiseRobustTest(Latency ~ Factor.int, data = exp2, est = "mom", nboot = 5000, method = "fdr")
PTLat

1.4) Draw graph of non-transformed data

LatInt <- ddply(exp2, .(Dose, Phenotype), summarise, val = mean(Latency))
ggplot(exp2, aes(x = factor(Dose), y = Latency, colour = Phenotype)) + geom_boxplot(outlier.shape = NA) + geom_point(data = LatInt, aes(y = val, group = Phenotype), position = position_dodge(width = 0.75)) + geom_line(data = LatInt, aes(y = val, group = Phenotype), position = position_dodge(width = 0.75)) + geom_jitter(position = "dodge", alpha = 0.3) + labs(x = "Dose", y = "Latency to first display (s)", color = "Phenotype")

1.5) Draw graph of Huber M-estimators and CIs

SumLat = groupwiseHuber(data = exp2copy, group = c("Phenotype", "Dose"), var = "Latency", conf.level = 0.95, conf.type="wald")
ggplot(SumLat, aes(x=Dose, y = M.Huber, color = Phenotype)) + geom_errorbar(aes(ymin=lower.ci, ymax=upper.ci), width = 0.2, size = 0.7, position = position_dodge(.2)) + geom_point(position = position_dodge(.2)) + geom_line(data = SumLat, aes(y = M.Huber, group = Phenotype), position = position_dodge(.2))

2. Display duration data 2.1) Run two-way ANOVAs on M-estimators

pbad2way(Dur.Display ~ Phenotype + Dose + Phenotype:Dose, data = exp2, est="mom", nboot = 5000)

2.2) Produce post-hoc tests for main effects

postDur.Display <- mcp2a(Dur.Display ~ Phenotype + Dose + Phenotype:Dose, data = exp2, est = "mom", nboot = 5000)
postDur.Display$contrasts
postDur.Display

2.3) Produce post-hoc tests for interaction effect

exp2$Factor.int <- interaction(exp2$Dose, exp2$Phenotype)
exp2$Factor.int <- factor(exp2$Factor.int, levels = c("0 mg/kg.LOF", "2.5 mg/kg.LOF", "5.0 mg/kg.LOF", "0 mg/kg.LEO", "2.5 mg/kg.LEO", "5.0 mg/kg.LEO"))
headTail(exp2)
PTDur.Display <- pairwiseRobustTest(Dur.Display ~ Factor.int, data = exp2, est = "mom", nboot = 5000, method = "fdr")
PTDur.Display

2.4) Draw graph of non-transformed data

Dur.DisplayInt <- ddply(exp2, .(Dose, Phenotype), summarise, val = mean(Dur.Display))
ggplot(exp2, aes(x = factor(Dose), y = Dur.Display, colour = Phenotype)) + geom_boxplot(outlier.shape = NA) + geom_point(data = Dur.DisplayInt, aes(y = val, group = Phenotype), position = position_dodge(width = 0.75)) + geom_line(data = Dur.DisplayInt, aes(y = val, group = Phenotype), position = position_dodge(width = 0.75)) + geom_jitter(position = "dodge", alpha = 0.3) + labs(x = "Dose", y = "Display duration (s)", color = "Phenotype")

2.5) Draw graph of Huber M-estimators and CIs

SumDur.Display = groupwiseHuber(data = exp2copy, group = c("Phenotype", "Dose"), var = "Dur.Display", conf.level = 0.95, conf.type="wald")
ggplot(SumDur.Display, aes(x=Dose, y = M.Huber, color = Phenotype)) + geom_errorbar(aes(ymin=lower.ci, ymax=upper.ci), width = 0.2, size = 0.7, position = position_dodge(.2)) + geom_point(position = position_dodge(.2)) + geom_line(data = SumDur.Display, aes(y = M.Huber, group = Phenotype), position = position_dodge(.2))

3. Display frequency data 3.1) Run two-way ANOVAs on M-estimators

pbad2way(N.Display ~ Phenotype + Dose + Phenotype:Dose, data = exp2, est="mom", nboot = 5000)

3.2) Produce post-hoc tests for main effects

postN.Display <- mcp2a(N.Display ~ Phenotype + Dose + Phenotype:Dose, data = exp2, est = "mom", nboot = 5000)
postN.Display$contrasts
postN.Display

3.3) Produce post-hoc tests for interaction effect

exp2$Factor.int <- interaction(exp2$Dose, exp2$Phenotype)
exp2$Factor.int <- factor(exp2$Factor.int, levels = c("0 mg/kg.LOF", "2.5 mg/kg.LOF", "5.0 mg/kg.LOF", "0 mg/kg.LEO", "2.5 mg/kg.LEO", "5.0 mg/kg.LEO"))
headTail(exp2)
PTN.Display <- pairwiseRobustTest(N.Display ~ Factor.int, data = exp2, est = "mom", nboot = 5000, method = "fdr")
PTN.Display

3.4) Draw graph of non-transformed data

N.DisplayInt <- ddply(exp2, .(Dose, Phenotype), summarise, val = mean(N.Display))
ggplot(exp2, aes(x = factor(Dose), y = N.Display, colour = Phenotype)) + geom_boxplot(outlier.shape = NA) + geom_point(data = N.DisplayInt, aes(y = val, group = Phenotype), position = position_dodge(width = 0.75)) + geom_line(data = N.DisplayInt, aes(y = val, group = Phenotype), position = position_dodge(width = 0.75)) + geom_jitter(position = "dodge", alpha = 0.3) + labs(x = "Dose", y = "Display frequency (N)", color = "Phenotype")

3.5) Draw graph of Huber M-estimators and CIs

SumN.Display = groupwiseHuber(data = exp2copy, group = c("Phenotype", "Dose"), var = "N.Display", conf.level = 0.95, conf.type="wald")
ggplot(SumN.Display, aes(x=Dose, y = M.Huber, color = Phenotype)) + geom_errorbar(aes(ymin=lower.ci, ymax=upper.ci), width = 0.2, size = 0.7, position = position_dodge(.2)) + geom_point(position = position_dodge(.2)) + geom_line(data = SumN.Display, aes(y = M.Huber, group = Phenotype), position = position_dodge(.2))

4. Time near mirror data 4.1) Run two-way ANOVAs on M-estimators

pbad2way(T.Mirror ~ Phenotype + Dose + Phenotype:Dose, data = exp2, est="mom", nboot = 5000)

4.2) Produce post-hoc tests for main effects

postT.Mirror <- mcp2a(T.Mirror ~ Phenotype + Dose + Phenotype:Dose, data = exp2, est = "mom", nboot = 5000)
postT.Mirror$contrasts
postT.Mirror

4.3) Produce post-hoc tests for interaction effect

exp2$Factor.int <- interaction(exp2$Dose, exp2$Phenotype)
exp2$Factor.int <- factor(exp2$Factor.int, levels = c("0 mg/kg.LOF", "2.5 mg/kg.LOF", "5.0 mg/kg.LOF", "0 mg/kg.LEO", "2.5 mg/kg.LEO", "5.0 mg/kg.LEO"))
headTail(exp2)
PTT.Mirror <- pairwiseRobustTest(T.Mirror ~ Factor.int, data = exp2, est = "mom", nboot = 5000, method = "fdr")
PTT.Mirror

4.4) Draw graph of non-transformed data

T.MirrorInt <- ddply(exp2, .(Dose, Phenotype), summarise, val = mean(T.Mirror))
ggplot(exp2, aes(x = factor(Dose), y = T.Mirror, colour = Phenotype)) + geom_boxplot(outlier.shape = NA) + geom_point(data = T.MirrorInt, aes(y = val, group = Phenotype), position = position_dodge(width = 0.75)) + geom_line(data = T.MirrorInt, aes(y = val, group = Phenotype), position = position_dodge(width = 0.75)) + geom_jitter(position = "dodge", alpha = 0.3) + labs(x = "Dose", y = "Time near mirror (s)", color = "Phenotype")

4.5) Draw graph of Huber M-estimators and CIs

SumT.Mirror = groupwiseHuber(data = exp2copy, group = c("Phenotype", "Dose"), var = "T.Mirror", conf.level = 0.95, conf.type="wald")
ggplot(SumT.Mirror, aes(x=Dose, y = M.Huber, color = Phenotype)) + geom_errorbar(aes(ymin=lower.ci, ymax=upper.ci), width = 0.2, size = 0.7, position = position_dodge(.2)) + geom_point(position = position_dodge(.2)) + geom_line(data = SumT.Mirror, aes(y = M.Huber, group = Phenotype), position = position_dodge(.2))

5. Locomotion data 5.1) Run two-way ANOVAs on M-estimators

pbad2way(Quad ~ Phenotype + Dose + Phenotype:Dose, data = exp2, est="mom", nboot = 5000)

5.2) Produce post-hoc tests for main effects

postQuad <- mcp2a(Quad ~ Phenotype + Dose + Phenotype:Dose, data = exp2, est = "mom", nboot = 5000)
postQuad$contrasts
postQuad

5.3) Produce post-hoc tests for interaction effect

exp2$Factor.int <- interaction(exp2$Dose, exp2$Phenotype)
exp2$Factor.int <- factor(exp2$Factor.int, levels = c("0 mg/kg.LOF", "2.5 mg/kg.LOF", "5.0 mg/kg.LOF", "0 mg/kg.LEO", "2.5 mg/kg.LEO", "5.0 mg/kg.LEO"))
headTail(exp2)
PTQuad <- pairwiseRobustTest(Quad ~ Factor.int, data = exp2, est = "mom", nboot = 5000, method = "fdr")
PTQuad

4.4) Draw graph of non-transformed data

QuadInt <- ddply(exp2, .(Dose, Phenotype), summarise, val = mean(Quad))
ggplot(exp2, aes(x = factor(Dose), y = Quad, colour = Phenotype)) + geom_boxplot(outlier.shape = NA) + geom_point(data = QuadInt, aes(y = val, group = Phenotype), position = position_dodge(width = 0.75)) + geom_line(data = QuadInt, aes(y = val, group = Phenotype), position = position_dodge(width = 0.75)) + geom_jitter(position = "dodge", alpha = 0.3) + labs(x = "Dose", y = "Number of squares crossed (N)", color = "Phenotype")

4.5) Draw graph of Huber M-estimators and CIs

SumQuad = groupwiseHuber(data = exp2copy, group = c("Phenotype", "Dose"), var = "Quad", conf.level = 0.95, conf.type="wald")
ggplot(SumQuad, aes(x=Dose, y = M.Huber, color = Phenotype)) + geom_errorbar(aes(ymin=lower.ci, ymax=upper.ci), width = 0.2, size = 0.7, position = position_dodge(.2)) + geom_point(position = position_dodge(.2)) + geom_line(data = SumQuad, aes(y = M.Huber, group = Phenotype), position = position_dodge(.2))