Read the data and transpose it, meaning the variables become columns and the observations become rows. This is the recommended format for statistical data analysis.
data %>%group_by(Tiempo, Calor, Congelado, Tratamiento) %>%count() %>%kable()
Table 1: Number of observations by treatment levels.
Tiempo
Calor
Congelado
Tratamiento
n
0
-
No
Liofilizado
5
0
60
No
Q1
4
0
75
No
Q2
5
0
No
No
Fresco
4
3
-
No
Liofilizado
5
3
60
-20
Q1C
5
3
60
-80
Q1U
5
3
75
-20
Q2C
4
3
75
-80
Q2C
5
3
No
-20
Congelado
5
3
No
-80
Ultracongelado
5
6
-
No
Liofilizado
3
6
60
-20
Q1C
5
6
60
-80
Q1U
5
6
75
-20
Q2C
5
6
75
-80
Q2C
5
6
No
-20
Congelado
5
6
No
-80
Ultraongelado
4
12
-
No
Liofilizado
5
12
60
-20
Q1C
5
12
60
-80
Q1U
4
12
75
-20
Q2C
4
12
75
-80
Q2C
5
12
No
-20
Congelado
5
12
No
-80
Ultraongelado
4
Review the data using the skim function to gain a general understanding of the experimental design, response variables, and factors.
Based on the data summary provided by the skim function, the dataset consists of 87 rows and 39 columns. The variables include both categorical and numerical variables, with 7 categorical variables and 32 numerical variables, mosty fatty acids.
Upon reviewing the categorical variables, it is noted that the variable “Tiempo” has 3 unique values, “Calor” has 4 unique values, “Congelado” has 3 unique values, “Tratamiento” has 10 unique values, “Réplica” has 5 unique values, “Etiqueta original” has 87 unique values, and “Código” has 87 unique values.
Mean and standard deviations are recorded for the different fats and fatty acids, along with statistics such as the minimum value, the 25th percentile, the median, the 75th percentile, and the maximum value for each variable.
Figure 1: Boxplot of fatty acids across different experimental times, categorised by calor levels. The y-axis is presented on a logarithmic scale.
Code
lines_plot_func(factor = Congelado)
Figure 2: Boxplot of fatty acids across different experimental times, categorised by Congelado levels. The y-axis is presented on a logarithmic scale.
Code
lines_plot_func(factor = Tratamiento)
Figure 3: Boxplot of fatty acids across different experimental times, categorised by Treatment levels. The y-axis is presented on a logarithmic scale.
3 Multivariate analysis
Multivariate analyses were conducted to describe patterns and identify drivers of changes in fatty acid composition using the vegan R library (Oksanen et al. 2012). Analyses were conducted based on Bray-Curtis dissimilarities of the log10 (x + 1) transformed data. Non-metric multidimensional scaling was used to visualise patterns in the multivariate data independent of predictor variables.
The differences in lipid profiles between Treatment, Time, Congelado, and Calor were tested using distance-based permutational multivariate multiple regression using 999 permutations.
Table 3: Results of PERMANOVA testing the effects of Calor, Congelado, Tratamiento and their interaction with Tiempo based on Bray-Curis dissimmilarities of the log transformed fatty acid data.
Term
Df
SS
R2
F
P
Calor
3
0.078
0.246
24.076
0.001
Congelado
2
0.014
0.045
6.609
0.001
Tratamiento
5
0.016
0.052
3.058
0.001
Tiempo
3
0.075
0.237
23.237
0.001
Calor:Tiempo
6
0.024
0.076
3.734
0.001
Congelado:Tiempo
2
0.005
0.017
2.456
0.019
Tratamiento:Tiempo
3
0.006
0.018
1.813
0.033
Residual
91
0.098
0.309
Total
115
0.316
1.000
4 Univariate analyses
Differences in fatty acid composition, including total saturated, total monounsaturated, PUFA - omega 6, PUFA - omega 3, and the omega 3 to omega 6 ratio, were assessed using linear mixed models. The fixed effects included Calor, Congelado, Tratamiento, and Tiempo. Sample ID was incorporated as a random effect to address the repeated measures experimental design.
Table 4: Summary of linear mixed models testing the effects Calor, Congelado, Tratamiento, and Tiempo of selected fatty acid variables. Sample ID was incorporated as a random effect to address the repeated measures experimental design.
