-
Notifications
You must be signed in to change notification settings - Fork 2
3 Output
When the user performs a data processing involving statistical methods (e.g. a differential analysis) DIMet generates tabular files that are explained in this section, both for the univariate and the bi-variate analyses.
DIMet computes the user defined analysis and generates tabular delimited files that have standardized columns:
Column name | Description/Interpretation |
---|---|
log2FC | log-transformed fold changes base 2. |
pvalue | computed by the chosen statistical test |
padj | adjusted p-values obtained by the chosen multiple correction method. |
distance/span | the normalized distance between the intervals of the two groups |
FC | fold changes between geometric means |
count_nan_samples_group1 | number of values missing in group 1 of samples. |
count_nan_samples_group2 | number of values missing in group 2 of samples. |
distance | numerator in the distance_over_span ratio |
span_allsamples | denominator in the distance_over_span ratio |
compartment | extracellular, cellular, or specific subcellular compartment |
gmean_1 | geometric mean of values from samples in group 1 |
gmean_2 | geometric mean of the values from samples in group 2 |
Special cases:
- The multi-group comparison only generates the statistic, pvalue and padj columns.
- If the user selects the Fitting of a distribution to the z-scores (abbreviation
disfit
) to test for the differential expression (whether in the pair-wise or time-course mode), the multiple correction method is senseless, so it is not applied. In this case the output does not report padj, and the significance is directly reported with the pvalue column.
The output table of the bi-variate analysis, performed with via Pearson’s correlation test, contains the following columns:
Column name | Description/Interpretation |
---|---|
correlation_coefficient | The correlation_coefficient is a value between -1 and 1 |
pvalue | Computed by the correlation test |
padj | The adjusted p-values obtained by the chosen multiple correction method |
gmean_arr_1 | For each metabolite, the array of the geometric means that correspond to the first variable (first condition or first time-point) |
gmean_arr_2 | For each metabolite, the array of the geometric means that correspond to the second variable (second condition or second time-point) |
The correlation_coefficient is a float between -1 and 1. Please read here for more information.
The output file name (click to show/hide)
The name of each file of the univariate analysis is the string formed by:- Input file name
- Compartment abbreviation
- Groups that were compared (does not apply in the multi-group analysis as all groups are considered)
- Abbreviation of the statistical test that was applied.
Example: Abundances--cell–drug-0h-vehicle-0h-ranksum.tsv
indicates a pair-wise differential analysis that compared group ‘drug 0h’ against ‘vehicle 0h’ group.
The name of each file of the bi-variate analysis includes:
- Input file name
- Compartment abbreviation
-
MDV
if the entire MDV profiles were compared - The groups compared:
- the two time-points, followed by the condition, or
- the two conditions, followed by the time point
- the statistical test that was applied.
Example: CorrectedIsotopologues_5--med--MDV-T2h-T0--Control-pearson.tsv
indicates that the MDV profiles were compared between time-points T2h and T0, of the Control condition, using the Pearson's test.
Note
In all the output files of the bivariate and univariate analyses, the significance is informed by the adjusted p-value (padj)