Ech.can_quistes

R packages

##
library(arm)
library(ggbreak)
library(scales)
library(tidyverse)
library(googlesheets4)
library(googledrive)
library(plotly)
library(lme4)
library(fitdistrplus)
library(goft)
library(data.table)

Data

ss= "https://docs.google.com/spreadsheets/d/1dr6LCQJevHdeS08zrdkbeONfhxU7etsojqVvBU7ryfs/edit?usp=sharing"
hoja = 1
rango = "A1:N193"

Descriptive

Proportion of infested pigs

Number of Cysts

Types of cysts

p <- ggplot(data=df,
            aes(x= Host.sex, fill= cyst.type) )
p <- p + geom_bar( stat = "count", color="white") +
  scale_fill_brewer(palette = "Blues") +
  labs(title="Types of cysts and sex of hosts", y= "n")
ggplotly(p)

## Fertile quistes
p <- ggplot(data=df,
            aes(x= Host.sex, fill= as.factor(fertile.bin)))
p <- p + geom_bar(stat = "count", position = "stack", color="white") +
  labs(title = "Fertile cysts") +
  scale_fill_brewer(palette="Blues") +
  labs(title="Fertile cysts and sex of hosts", y= "n")
ggplotly(p)

##
p <- ggplot(data=df,
            aes( x= size_cm) )
p <- p + geom_histogram(bins=20, binwidth = 0.2, color="white", fill="darkgrey") +
  scale_fill_brewer(palette="Blues") +
  labs(title = "Hisrogram for the size of a cyst (cm)")
ggplotly(p)

Associations

Generalized Logistic Model

Host individual's Sex & Cyst size

Call:
glm(formula = fertile.bin ~ size_cm * Host.sex, family = "binomial", 
    data = df)

Coefficients:
                      Estimate Std. Error z value Pr(>|z|)    
(Intercept)          -2.941051   0.789402  -3.726 0.000195 ***
size_cm               0.431272   0.390205   1.105 0.269054    
Host.sexMale          1.275366   0.892428   1.429 0.152976    
size_cm:Host.sexMale -0.001491   0.429234  -0.003 0.997228    
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

(Dispersion parameter for binomial family taken to be 1)

    Null deviance: 213.71  on 191  degrees of freedom
Residual deviance: 194.76  on 188  degrees of freedom
AIC: 202.76

Number of Fisher Scoring iterations: 5


Call:
glm(formula = fertile.bin ~ Host.sex + size_cm, family = "binomial", 
    data = df)

Coefficients:
             Estimate Std. Error z value Pr(>|z|)    
(Intercept)   -2.9391     0.5424  -5.418 6.02e-08 ***
Host.sexMale   1.2728     0.5165   2.464  0.01373 *  
size_cm        0.4300     0.1626   2.645  0.00817 ** 
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

(Dispersion parameter for binomial family taken to be 1)

    Null deviance: 213.71  on 191  degrees of freedom
Residual deviance: 194.76  on 189  degrees of freedom
AIC: 200.76

Number of Fisher Scoring iterations: 5

• Both the sex of the host individual (i.e., Males) and the size of the cyst itself are good predictors of cyst fertility (P= 0.01).

• The cysts collected from male pigs have 3.6 (CI95% 1.3-9.8) more times to be fertile than the ones collected from female pigs. This suggests some factors associated with the host individual explain, at least partially, the fertility of the cyst

Generalized Logistic Mixed Model [“host individual” as random effect]

Testing for Random effects at the host individual scale (Baseline glm Vs. baseline mixed-model )

[1] 160.6177
[1] 215.7116
[1] 4.155105e-14

• The random effect of the host individual is justified. The basal AIC (null model) with the random effect included is less (P< 0.001) than without including that term. This suggests there are additional factors at the individual scale (e.g., sex of the animal) possibly not previously measured or studied, that are associated with the probability of fertility.

Host's Sex + random Fx

Generalized linear mixed model fit by maximum likelihood (Laplace Approximation) ['glmerMod']
 Family: binomial  ( logit )
Formula: fertile.bin ~ (1 | code) + Host.sex
   Data: df

     AIC      BIC   logLik deviance df.resid 
   162.1    171.9    -78.0    156.1      189 

Scaled residuals: 
    Min      1Q  Median      3Q     Max 
-2.1186 -0.2690 -0.1283 -0.1030  3.7174 

Random effects:
 Groups Name        Variance Std.Dev.
 code   (Intercept) 5.794    2.407   
Number of obs: 192, groups:  code, 39

Fixed effects:
             Estimate Std. Error z value Pr(>|z|)   
(Intercept)   -4.1193     1.3264  -3.106   0.0019 **
Host.sexMale   0.8578     1.1719   0.732   0.4642   
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Correlation of Fixed Effects:
            (Intr)
Host.sexMal -0.561

Cyst size + random Fx

Generalized linear mixed model fit by maximum likelihood (Laplace Approximation) ['glmerMod']
 Family: binomial  ( logit )
Formula: fertile.bin ~ (1 | code) + size_cm
   Data: df

     AIC      BIC   logLik deviance df.resid 
   158.3    168.1    -76.2    152.3      189 

Scaled residuals: 
    Min      1Q  Median      3Q     Max 
-2.3304 -0.2820 -0.1102 -0.0700  3.1833 

Random effects:
 Groups Name        Variance Std.Dev.
 code   (Intercept) 6.386    2.527   
Number of obs: 192, groups:  code, 39

Fixed effects:
            Estimate Std. Error z value Pr(>|z|)    
(Intercept)  -4.8529     1.3814  -3.513 0.000443 ***
size_cm       0.6644     0.3358   1.979 0.047857 *  
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Correlation of Fixed Effects:
        (Intr)
size_cm -0.592

• After including the animal identifier as random effect, only the the size of the cyst (unlike the Sex of the host) is an suitable candidate to explain the probability of fertility of a cyst (P=0.047).

Cyst fertility~ Cyst's Size [Unimodal]

Call:
glm(formula = fertile.bin ~ size_cm + I(size_cm^2), family = "binomial", 
    data = df)

Coefficients:
             Estimate Std. Error z value Pr(>|z|)    
(Intercept)   -5.5840     1.0450  -5.343 9.12e-08 ***
size_cm        4.0952     0.9329   4.390 1.14e-05 ***
I(size_cm^2)  -0.7436     0.1927  -3.859 0.000114 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

(Dispersion parameter for binomial family taken to be 1)

    Null deviance: 213.71  on 191  degrees of freedom
Residual deviance: 177.68  on 189  degrees of freedom
AIC: 183.68

Number of Fisher Scoring iterations: 5

Optimum size for Cyst Fertility probability

size_cm 
   2.75 

GLM model

• Optimum size= 3.15cm.
• Highest probability range for Cyst fertility (Tolerance)= 2.09cm-4.2cm.

Unimodal Mixed Model - GLMM (Random Effects: Host individual)

Generalized linear mixed model fit by maximum likelihood (Laplace Approximation) ['glmerMod']
 Family: binomial  ( logit )
Formula: fertile.bin ~ (1 | code) + size_cm + I(size_cm^2)
   Data: df

     AIC      BIC   logLik deviance df.resid 
   157.2    170.2    -74.6    149.2      188 

Scaled residuals: 
     Min       1Q   Median       3Q      Max 
-2.05398 -0.28319 -0.13012 -0.04482  3.10260 

Random effects:
 Groups Name        Variance Std.Dev.
 code   (Intercept) 4.422    2.103   
Number of obs: 192, groups:  code, 39

Fixed effects:
             Estimate Std. Error z value Pr(>|z|)    
(Intercept)   -6.2826     1.6063  -3.911 9.18e-05 ***
size_cm        2.8215     1.3012   2.168   0.0301 *  
I(size_cm^2)  -0.4484     0.2559  -1.752   0.0797 .  
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Correlation of Fixed Effects:
            (Intr) siz_cm
size_cm     -0.783       
I(siz_cm^2)  0.687 -0.968

size_cm 
   3.15 

• Optimum size= 3.15cm.
• Highest probability range for Cyst fertility (Tolerance)= 2.09cm-4.2cm.