To present the base functionalities of MetAlyzer, we use a data set
published by Gegner
et al. 2022. This data set was analyzed using the MxP®Quant
500 kit and inlcudes 6 different extraction methods (1 to 6) to quantify
630 metabolites in triplicates for 4 tissue types (Drosophila, Mouse
Liver, Mouse Kidney, Zebrafish Liver). It is included in the MetAlyzer
package and can be accessed via example_extraction_data().
MetAlyzer_dataset reads the data from Excel into R and
stores it as a SummarizedExperiment (SE).
The SE includes assay information with concentration values and
quantification status, while storing metabolites and their respective
classes in the row data. It is important to note that there are 630
metabolites, and 232 metabolite indicators calculated using the MetIDQ™
software. Sample-wise metadata is stored in the column data. Additional
data such as the file path (‘file_path’), the Excel sheet
number (‘sheet_index’), a list of colors for each
quantification status (status_list), as well as a tibble of
aggregated concentration data and quantification status for downstream
analysis and visualizations (‘aggregated_data’) are stored in
the SE’s metadata.
Summaries of concentration values and quantification status are
automatically called during data loading, but can also manually be
called using summarizeConcValues(metalyzer_se) and
summarizeQuantData(metalyzer_se). The function
summarizeConcValues prints the quantiles and the amount /
percentage of NAs in the concentration data. The function
summarizeQuantData gives an overview of the quantification
status of the metabolites. E.g 34.16% of metabolites where below the
Limit of Detection.
The various elements within the SummarizedExperiment (SE) can be
accessed in a manner consistent with a typical SE.
concentration_values <- assays(metalyzer_se)$conc_values
head(concentration_values, c(5, 5))
#> 9 10 11 12 13
#> C0 203.000 86.800 246.000 198.000 369.000
#> C2 29.500 15.800 34.600 39.200 42.200
#> C3 39.200 9.290 49.900 20.100 50.000
#> C3-DC (C4-OH) 0.057 0.055 0.076 0.679 0.071
#> C3-OH 0.106 0.132 0.116 0.145 0.170
quantification_status <- assays(metalyzer_se)$quant_status
head(quantification_status, c(5, 5))
#> 9 10 11 12 13
#> C0 "Valid" "Valid" "Valid" "Valid" "Valid"
#> C2 "Valid" "Valid" "Valid" "Valid" "Valid"
#> C3 "Valid" "Valid" "Valid" "Valid" "Valid"
#> C3-DC (C4-OH) "LOD" "LOD" "LOD" "Valid" "LOD"
#> C3-OH "LOD" "LOD" "LOD" "LOD" "LOD"
Analysis and visualization of treatment data
MetAlyzer also includes functions to calculate and visualize the log2
fold change of data with paired effects. For this, we load a dataset
with control and mutated samples analyzed using the MxP®Quant 500 XL
kit. The XL kit covers 332 additional metabolites and 6 additional
metabolic classes compared to the MxP®Quant 500 kit.
fpath <- example_mutation_data_xl()
metalyzer_se <- MetAlyzer_dataset(file_path = fpath)
#>
#>
#> _____ ______ _______ _________ ________ ___ ___ ___ ________ _______ ________
#> |\ _ \ _ \|\ ___ \|\___ ___\\ __ \|\ \ |\ \ / /|\_____ \|\ ___ \ |\ __ \
#> \ \ \\\__\ \ \ \ __/\|___ \ \_\ \ \|\ \ \ \ \ \ \/ / /\|___/ /\ \ __/|\ \ \|\ \
#> \ \ \\|__| \ \ \ \_|/__ \ \ \ \ \ __ \ \ \ \ \ / / / / /\ \ \_|/_\ \ _ _\
#> \ \ \ \ \ \ \ \_|\ \ \ \ \ \ \ \ \ \ \ \____ \/ / / / /_/__\ \ \_|\ \ \ \\ \|
#> \ \__\ \ \__\ \_______\ \ \__\ \ \__\ \__\ \_______\__/ / / |\________\ \_______\ \__\\ _\
#> \|__| \|__|\|_______| \|__| \|__|\|__|\|_______|\____/ / \|_______|\|_______|\|__|\|__|
#> \|____|/
#>
#>
#> Info: Reading color code "FFCBD2D7" as "#CBD2D7"
#> Info: Reading color code "FF7FB2C5" as "#7FB2C5"
#> Info: Reading color code "FFCBD2D7" as "#CBD2D7"
#> Info: Reading color code "FFB9DE83" as "#B9DE83"
#> Info: Reading color code "FFB9DE83" as "#B9DE83"
#> Info: Reading color code "FFA28BA3" as "#A28BA3"
#> Info: Reading color code "FFA28BA3" as "#A28BA3"
#> Info: Reading color code "FFB2D1DC" as "#B2D1DC"
#> Info: Reading color code "FFB2D1DC" as "#B2D1DC"
#> Info: Reading color code "FF7FB2C5" as "#7FB2C5"
#>
#> Measured concentration values:
#> ------------------------------
#> 0% 25% 50% 75% 100%
#> 0.00 1.63 7.77 43.07 139200.40
#>
#> NAs: 155 (0.85%)
#>
#>
#> Measured quantification status:
#> -------------------------------
#> Valid: 14280 (77.85%)
#> LOQ: 819 (4.47%)
#> LOD: 3243 (17.68%)
#> NAs: 0 (0%)
metalyzer_se
#> class: SummarizedExperiment
#> dim: 1019 18
#> metadata(4): file_path sheet_index status_list aggregated_data
#> assays(2): conc_values quant_status
#> rownames(1019): C0 C2 ... TG 22:6_34:3 Choline
#> rowData names(1): metabolic_classes
#> colnames(18): 32 33 ... 48 49
#> colData names(11): Plate bar code Sample bar code ... Org. info
#> Measurement time
Again, we want to remove the metabolism indicators.
metalyzer_se <- filterMetabolites(metalyzer_se, drop_metabolites = "Metabolism Indicators")
#> Info: No metabolites were removed!
metalyzer_se
#> class: SummarizedExperiment
#> dim: 1019 18
#> metadata(4): file_path sheet_index status_list aggregated_data
#> assays(2): conc_values quant_status
#> rownames(1019): C0 C2 ... TG 22:6_34:3 Choline
#> rowData names(1): metabolic_classes
#> colnames(18): 32 33 ... 48 49
#> colData names(11): Plate bar code Sample bar code ... Org. info
#> Measurement time
In this data set, the column “Sample Description” holds information
about the wether a sample is a wild type or a mutant.
meta_data <- colData(metalyzer_se)
meta_data$`Sample Description`
#> [1] "Mutant" "Mutant" "Mutant" "Mutant" "Mutant" "Mutant" "Mutant"
#> [8] "Mutant" "Mutant" "Control" "Control" "Control" "Control" "Control"
#> [15] "Control" "Control" "Control" "Control"
To determine the direction of the effect during log2 fold change
calculation, the information is converted into a factor and saved as a
new column “Control_Mutant”. The log2 fold change will now be calculated
from “Control” to “Mutant”.
control_mutant <- factor(colData(metalyzer_se)$`Sample Description`, levels = c("Control", "Mutant"))
metalyzer_se <- updateMetaData(metalyzer_se, Control_Mutant = control_mutant)
meta_data <- colData(metalyzer_se)
meta_data$Control_Mutant
#> [1] Mutant Mutant Mutant Mutant Mutant Mutant Mutant Mutant Mutant
#> [10] Control Control Control Control Control Control Control Control Control
#> Levels: Control Mutant
In order to avoid NAs during log2 transformation the same imputation
of NAs or zero values is applied as for calculate_anova. To
impute NAs, the impute_NA argument can be set to TRUE. The
argument impute_perc_of_min specifies the percentage of the
minimum concentration value that is used for imputation.
metalyzer_se <- calculate_log2FC(
metalyzer_se,
categorical = "Control_Mutant",
impute_perc_of_min = 0.2,
impute_NA = TRUE
)
#> Info: Imputing concentrations (groupwise: Metabolite) with 20% of the minimal positive value... finished!
#> Info: Calculating log2 fold change from Mutant to Control (column: Control_Mutant).
#> Info: Calculating log2 fold change groupwise (Metabolite) using a linear model... finished!
In addition to accessing the log2 fold change data through
metadata(metalyzer_se)$log2FC, it can be directly accessed
using log2FC.
log2FC(metalyzer_se)
#> # A tibble: 1,019 × 5
#> Metabolite Class log2FC pval qval
#> <fct> <fct> <dbl> <dbl> <dbl>
#> 1 C0 Acylcarnitines -0.0953 0.694 0.838
#> 2 C2 Acylcarnitines 0.323 0.431 0.643
#> 3 C3 Acylcarnitines -0.351 0.219 0.488
#> 4 C3-DC (C4-OH) Acylcarnitines 0.315 0.420 0.633
#> 5 C3-OH Acylcarnitines 0.593 0.460 0.667
#> 6 C3:1 Acylcarnitines 0.203 0.781 0.894
#> 7 C4 Acylcarnitines 0.240 0.185 0.462
#> 8 C4:1 Acylcarnitines 1.11 0.365 0.585
#> 9 C5 Acylcarnitines 0.115 0.628 0.791
#> 10 C5-DC (C6-OH) Acylcarnitines 0.342 0.0498 0.272
#> # ℹ 1,009 more rows
The log2 fold change between the wild type and the mutant can be
visualized with a volcano plot:
log2fc_vulcano <- plot_log2FC(
metalyzer_se,
hide_labels_for = rownames(rowData(metalyzer_se)),
vulcano=TRUE
)
log2fc_vulcano
as well as in a scatter plot categorized by metabolic classes:
log2fc_by_class <- plot_log2FC(
metalyzer_se,
hide_labels_for = rownames(rowData(metalyzer_se)),
vulcano=FALSE
)
log2fc_by_class
MetAlyzer also provides a visualization of the log2 fold change
across different pathways, in order to identify groups of metabolites
impacted by the effect.
log2fc_network <- plot_network(
metalyzer_se,
q_value=0.05,
metabolite_text_size=2,
connection_width=0.75,
pathway_text_size=4,
pathway_width=4,
scale_colors = c("green", "black", "magenta")
)
#> Warning: Removing 5 invalid nodes.
log2fc_network
