metabolite-class-subset-filtering
Summary
Filter a metabolomics summary statistics dataframe to isolate metabolites belonging to a specific compound class and meeting statistical significance thresholds (padj ≤ 0.05). This subset becomes the basis for downstream frequency counting and visualization of altered metabolites within that class.
When to use
When you have an omu_summary output dataframe with Class metadata and adjusted p-values (padj), and you need to focus downstream analysis (e.g., pie chart visualization, fold-change counting) on a specific metabolite class — such as organic acids, fatty acids, or amino acids — that shows significant alteration (padj ≤ 0.05) in your experiment.
When NOT to use
- Input dataframe does not contain both Class and padj columns — subsetting will fail or yield empty results.
- Your analysis goal is to examine all metabolites regardless of class or significance — filtering removes data and is unnecessary.
- The padj threshold or Class label does not match your biological question or experimental design.
Inputs
- omu_summary output dataframe (containing metabolite names, Class metadata, padj values, log2FoldChange)
Outputs
- Filtered dataframe (subset of omu_summary, rows matching Class and padj ≤ 0.05)
How to apply
Load the omu_summary output dataframe containing metabolite names, Class metadata, padj values, and log2FoldChange columns. Use base R subsetting (e.g., subset() or logical indexing) to filter rows where Class matches your target (e.g., 'Organic acids') AND padj ≤ 0.05. The resulting subset preserves all column structure and is ready for input to downstream omu functions such as count_fold_changes (with column='Class' and sig_threshold=0.05) to generate frequency tables of increased vs. decreased metabolites, or ra_table to compute relative abundance percentages for visualization.
Related tools
- omu_summary (Generates the input dataframe containing padj and log2FoldChange statistics for metabolites) — github.com/connor-reid-tiffany/Omu
- count_fold_changes (Consumes the filtered dataframe to generate frequency tables of increased vs. decreased metabolites by class) — github.com/connor-reid-tiffany/Omu
- ra_table (Transforms frequency data from the filtered subset into relative abundance (percentage) format for visualization) — github.com/connor-reid-tiffany/Omu
- R base subset / logical indexing (Core subsetting mechanism to filter rows by Class and padj criteria)
Examples
filtered_organicacids <- subset(omu_summary_df, Class == 'Organic acids' & padj <= 0.05)
Evaluation signals
- Verify output dataframe has fewer or equal rows than input; all retained rows have Class matching the target label and padj ≤ 0.05.
- Check that all columns from the input omu_summary output are preserved in the filtered subset (no unexpected column drops).
- Confirm the filtered dataframe is non-empty and contains at least one metabolite; if empty, the class label or threshold may not match the data distribution.
- Validate that downstream functions (count_fold_changes, ra_table, pie_chart) accept the filtered dataframe without errors and produce expected frequency/visualization outputs.
- Cross-check row counts and class labels before and after subsetting using summary() or table(subset_df$Class) to confirm filtering logic worked as intended.
Limitations
- Subsetting is case-sensitive; Class labels must match exactly (e.g., 'Organic acids' vs. 'organic acids' will yield different subsets).
- If no metabolites meet both the Class and padj ≤ 0.05 criteria, the output is an empty dataframe, which may cause downstream functions to fail or return no visualization.
- The padj threshold (0.05) is fixed in the workflow; if your analysis requires a different significance cutoff, manual adjustment is needed.
- Filtering on Class alone does not account for biological or statistical effect size; metabolites with padj ≤ 0.05 but small log2FoldChange may not be biologically meaningful.
Evidence
- [other] Subset the dataframe to rows where Class == 'Organic acids' and padj <= 0.05 using base R subset and indexing.: "Subset the dataframe to rows where Class == 'Organic acids' and padj <= 0.05 using base R subset and indexing."
- [other] Load the omu_summary output dataframe (containing metabolite names, Class metadata, padj values, and log2FoldChange).: "Load the omu_summary output dataframe (containing metabolite names, Class metadata, padj values, and log2FoldChange)."
- [other] Apply count_fold_changes with column='Class' and sig_threshold=0.05 to generate a frequency table counting increased vs. decreased metabolites.: "Apply count_fold_changes with column='Class' and sig_threshold=0.05 to generate a frequency table counting increased vs. decreased metabolites."
- [other] Omu supports two univariate statistical models, t test and anova, using the functions
omu_summary and anova_function respectively: "Omu supports two univariate statistical models, t test and anova, using the functions omu_summary and anova_function respectively"
1---2name: metabolite-class-subset-filtering3description: Use when when you have an omu_summary output dataframe with Class metadata and adjusted p-values (padj), and you need to focus downstream analysis (e.4license: CC-BY-4.05---67# metabolite-class-subset-filtering89## Summary1011Filter a metabolomics summary statistics dataframe to isolate metabolites belonging to a specific compound class and meeting statistical significance thresholds (padj ≤ 0.05). This subset becomes the basis for downstream frequency counting and visualization of altered metabolites within that class.1213## When to use1415When you have an omu_summary output dataframe with Class metadata and adjusted p-values (padj), and you need to focus downstream analysis (e.g., pie chart visualization, fold-change counting) on a specific metabolite class — such as organic acids, fatty acids, or amino acids — that shows significant alteration (padj ≤ 0.05) in your experiment.1617## When NOT to use1819- Input dataframe does not contain both Class and padj columns — subsetting will fail or yield empty results.20- Your analysis goal is to examine all metabolites regardless of class or significance — filtering removes data and is unnecessary.21- The padj threshold or Class label does not match your biological question or experimental design.2223## Inputs2425- omu_summary output dataframe (containing metabolite names, Class metadata, padj values, log2FoldChange)2627## Outputs2829- Filtered dataframe (subset of omu_summary, rows matching Class and padj ≤ 0.05)3031## How to apply3233Load the omu_summary output dataframe containing metabolite names, Class metadata, padj values, and log2FoldChange columns. Use base R subsetting (e.g., subset() or logical indexing) to filter rows where Class matches your target (e.g., 'Organic acids') AND padj ≤ 0.05. The resulting subset preserves all column structure and is ready for input to downstream omu functions such as count_fold_changes (with column='Class' and sig_threshold=0.05) to generate frequency tables of increased vs. decreased metabolites, or ra_table to compute relative abundance percentages for visualization.3435## Related tools3637- **omu_summary** (Generates the input dataframe containing padj and log2FoldChange statistics for metabolites) — github.com/connor-reid-tiffany/Omu38- **count_fold_changes** (Consumes the filtered dataframe to generate frequency tables of increased vs. decreased metabolites by class) — github.com/connor-reid-tiffany/Omu39- **ra_table** (Transforms frequency data from the filtered subset into relative abundance (percentage) format for visualization) — github.com/connor-reid-tiffany/Omu40- **R base subset / logical indexing** (Core subsetting mechanism to filter rows by Class and padj criteria)4142## Examples4344```45filtered_organicacids <- subset(omu_summary_df, Class == 'Organic acids' & padj <= 0.05)46```4748## Evaluation signals4950- Verify output dataframe has fewer or equal rows than input; all retained rows have Class matching the target label and padj ≤ 0.05.51- Check that all columns from the input omu_summary output are preserved in the filtered subset (no unexpected column drops).52- Confirm the filtered dataframe is non-empty and contains at least one metabolite; if empty, the class label or threshold may not match the data distribution.53- Validate that downstream functions (count_fold_changes, ra_table, pie_chart) accept the filtered dataframe without errors and produce expected frequency/visualization outputs.54- Cross-check row counts and class labels before and after subsetting using summary() or table(subset_df$Class) to confirm filtering logic worked as intended.5556## Limitations5758- Subsetting is case-sensitive; Class labels must match exactly (e.g., 'Organic acids' vs. 'organic acids' will yield different subsets).59- If no metabolites meet both the Class and padj ≤ 0.05 criteria, the output is an empty dataframe, which may cause downstream functions to fail or return no visualization.60- The padj threshold (0.05) is fixed in the workflow; if your analysis requires a different significance cutoff, manual adjustment is needed.61- Filtering on Class alone does not account for biological or statistical effect size; metabolites with padj ≤ 0.05 but small log2FoldChange may not be biologically meaningful.6263## Evidence6465- [other] Subset the dataframe to rows where Class == 'Organic acids' and padj <= 0.05 using base R subset and indexing.: "Subset the dataframe to rows where Class == 'Organic acids' and padj <= 0.05 using base R subset and indexing."66- [other] Load the omu_summary output dataframe (containing metabolite names, Class metadata, padj values, and log2FoldChange).: "Load the omu_summary output dataframe (containing metabolite names, Class metadata, padj values, and log2FoldChange)."67- [other] Apply count_fold_changes with column='Class' and sig_threshold=0.05 to generate a frequency table counting increased vs. decreased metabolites.: "Apply count_fold_changes with column='Class' and sig_threshold=0.05 to generate a frequency table counting increased vs. decreased metabolites."68- [other] Omu supports two univariate statistical models, t test and anova, using the functions ```omu_summary``` and ```anova_function``` respectively: "Omu supports two univariate statistical models, t test and anova, using the functions ```omu_summary``` and ```anova_function``` respectively"
HolobiomicsLab