dimet bivariate analysis

This module is part of DIMet: Computation of the correlation of entire MDV profiles, or the metabolite time course profiles (https://pypi.org/project/DIMet/).

DIMet bi-variate analysis performs the comparison of entire MDV profiles, with the user provided isotopologue proportions data. Moreover, when total abundances and/or mean enrichment are provided, the comparison of the metabolite time-course profiles is also computed. Specifically, three types of bi-variate comparisons are performed automatically:

• MDV profile comparison between two conditions
• MDV profile comparison between two consecutive time-points
• Metabolite (total abundances and/or mean enrichment) time course profiles comparison between two conditions

For all these three types of bi-variate comparison, the statistical test that is applied is the Pearson's correlation test.

To note, MDV (Mass Distribution Vector) are obtained automatically by the tool, using the isotopologue proportions.

Input data files

This tool requires (at max.) 4 tab-delimited .csv files as inputs. There are two types of files:

• The measures' (or quantifications') files, that can be of 3 types.
• The metadata, a unique file with the description of the samples in your measures' files. This is compulsory.

For running DIMet bivariate_analysis you need at least one file of measures:

• The isotopologue proportions file
• The total abundances (of the metabolites) file
• The mean enrichment or labelled fractional contributions

and one metadata file, WHICH IS COMPULSORY, see section Metadata File Information.

Measures files

The measures files must be organized as matrices:

• The first column must contain Metabolite IDs that are unique (not repeated) within the file.
• The rest of the columns correspond to the samples
• The rows correspond to the metabolites
• The values must be tab separated, with the first row containing the sample/column labels.

See the following examples of measures files:

Example - Metabolites abundances:

ID	MCF001089_TD01	MCF001089_TD02	MCF001089_TD03	MCF001089_TD04	MCF001089_TD05	MCF001089_TD06
2_3-PG	8698823.9926	10718737.7217	10724373.9	8536484.5	22060650	28898956
2-OHGLu	36924336	424336	92060650	45165	84951950	965165051
Glc6P	2310	2142	2683	1683	012532068	1252172
Gly3P	399298	991656565	525195	6365231	89451625	4952651963
IsoCit	0	0	0	84915613	856236	954651610

Example - mean enrichment or labeled fractional contributions:

ID	MCF001089_TD01	MCF001089_TD02	MCF001089_TD03	MCF001089_TD04	MCF001089_TD05	MCF001089_TD06
2_3-PG	0.9711	0.968	0.9909	0.991	0.40	0.9
2-OHGLu	0.01719	0.0246	0.554	0.555	0.73	0.68
Glc6P	0.06	0.66	2683	0.06	2068	2172
Gly3P	0.06	0.06	0.06	1	5	3
IsoCit	0.06	1	0.49	0.36	6	10

Example - Isotopologue proportions:

ID	MCF001089_TD01	MCF001089_TD02	MCF001089_TD03	MCF001089_TD04	MCF001089_TD05	MCF001089_TD06
2_3-PG_m+0	0.023701408	0.026667837	0.003395407	0.05955	0.034383527	0.12
2_3-PG_m+1	0.0	0.0	0.0	0.0	0.4	0.12
2_3-PG_m+2	0.015379329	0.01506	0.017029723	0.35483229	0.54131313	0.743
2_3-PG_m+3	0.960919263	0.958268099	0.97957487	0.581310816	0.017029723	0.017
2-OHGLu_m+0	0.972778716	0.960016157	0.238843937	0.234383527	0.9998888	0.015064063
2-OHGLu_m+1	0.0	0.0	0.0	0.0	0.0001112	0.960919263

Metadata File Information

Provide a tab-separated file that has the names of the samples in the first column and one header row. Column names must be exactly in this order:

name_to_plot condition timepoint timenum compartment original_name

Example Metadata File:

name_to_plot	condition	timepoint	timenum	compartment	original_name
Control_cell_T0-1	Control	T0	0	cell	MCF001089_TD01
Control_cell_T0-2	Control	T0	0	cell	MCF001089_TD02
Control_cell_T0-3	Control	T0	0	cell	MCF001089_TD03
Tumoral_cell_T0-1	Tumoral	T0	0	cell	MCF001089_TD04
Tumoral_cell_T0-2	Tumoral	T0	0	cell	MCF001089_TD05
Tumoral_cell_T0-3	Tumoral	T0	0	cell	MCF001089_TD06
Tumoral_cell_T24-1	Tumoral	T24	24	cell	MCF001089_TD07
Tumoral_cell_T24-2	Tumoral	T24	24	cell	MCF001089_TD08
Tumoral_cell_T24-3	Tumoral	T24	24	cell	MCF001090_TD01
Control_med_T24-1	Control	T24	24	med	MCF001090_TD02
Control_med_T24-2	Control	T24	24	med	MCF001090_TD03
Tumoral_med_T24-1	Tumoral	T24	24	med	MCF001090_TD04
Tumoral_med_T24-2	Tumoral	T24	24	med	MCF001090_TD05
Control_med_T0-1	Control	T0	0	med	MCF001090_TD06
Tumoral_med_T0-1	Tumoral	T0	0	med	MCF001090_TD07
Tumoral_med_T0-2	Tumoral	T0	0	med	MCF001090_TD08

The column original_name must have the names of the samples as given in your data.

The column name_to_plot must have the names as you want them to be (or set identical to original_name if you prefer). To set names that are meaningful is a better choice, as we will take them to display the results.

The column timenum must contain only the numeric part of the timepoint, for example 2,0, 10, 100 (this means, without letters ("T", "t", "s", "h" etc) nor any other symbol). Make sure these time numbers are in the same units (but do not write the units here!).

The column compartment is an abbreviation, coined by you, for the compartments. This will be used for the results' files names: the longer the compartments names are, the longer the output files' names! Please pick short and clear abbreviations to fill this column.

Running the analysis

You can precise how you want your analysis to be executed, with the parameters:

• datatypes : the measures type(s) that you want to run.
• conditions: the two conditions to be compared in the bi-variate analysis. If 3 or more conditions are set, the tool automatically performs all the 1-to-1 condition comparisons. If you only have one condition in your data, select the condition, and see the Note at the end of this section.
• correction_method : one of the methods for multiple testing correction available in statsmodels library (bonferroni, fdr_bh, sidak, among others, see https://www.statsmodels.org/dev/generated/statsmodels.stats.multitest.multipletests.html).

Note: the tool automatically performs the comparison of MDV profile between two consecutive time-points. All the time-points are processed.

There exist hints on use that will guide you, next to the parameters.

For more information about the implemented statistical tests, please visit: https://github.com/cbib/DIMet/wiki/2-Statistical-tests

The output files are explained in https://github.com/cbib/DIMet/wiki/3-Output

Available data for testing

You can test our tool with the data from our manuscript https://zenodo.org/records/10579862 (the pertinent files for you are located in the subfolders inside the data folder). You can also use the minimal data examples from https://zenodo.org/records/10579891