Map omics data onto metabolic networks

Mapping Demo
Select the menu Omics->Mapping->New and click on the Demo button




First, select the BioSource on which you want to map your data. Then, click on the item "Mapping" of the menu "Mapping".

Input data

Omics data can be uploaded by two ways in MetExplore:

  • A text file with columns: By default, columns are separated by tabulations but the separator can be changed in the mapping form.
  • By copy paste data from Excel or text tabulated file: Copy (Ctrl-C) your data in your Excel/tabulated file and paste them with Ctrl-V directly in the mapping grid.
In both cases, the first column must be the features of the biological objects (see next section) which will allow to map them in the biological network. Following columns are facultative and correspond to numeric values in different conditions. If the first line of your data contains the names of these conditions, don't forget to check "Consider first row as header of columns".



Copy-paste your data from Excel files.

Selection of the biological objects to map

Mapping can be performed in all biological objects stored in a metabolic network (pathways, reactions, metabolites, enzymes, gene products and genes). Select the type of the biological objects you want to map in the "Object" menu in the mapping grid. Then, in the menu "Element" select the feature of the objects which will be used to identify them.
Each biological object can be mapped thanks to its name or its identifier.

For metabolites, two additional features are available for the mapping: the Inchi code and the monoisotopic mass. In the latter case, the user can indicate the allowed error in ppm in the menu below the mapping grid.

Mapping modes

One mapping mode

The mapping default mode is the "one mapping mode". In this mode, only one mapping is done. This mode is used to simply map a list of objects (one column) with associated single numerical values (two columns) or to compare the same network in different conditions (multiple columns). Mapping and pathway enrichment will be performed for all elements in the first column, irrespectively of the values in the different columns.

Multiple mapping mode

The "multi mapping" mode will produce one mapping for each column. In each mapping, the objects for which the value is different from NaN will be tagged as identified in the grid and pathway enrichment will be performed only on these objects.



Input file for a multiple mapping. Two mappings will be created: one for the Intestine condition and one for the Liver condition. In the Intestine condition, metabolites that will mapped as identified will be those that have a value different from NaN in the Intestine condition.

Mapping results displayed in grids

Results on the grid corresponding to the biological object mapped

In the data grid corresponding to the identified biological object, a main column by mapping is added. This main column is subdivided into several sub-columns. One indicates if the object has been identified and the others correspond to the values ​​indicated in the other columns of the input file, if they exist.

Propagation

Mapping on any biological object triggers an enrichment analysis of the metabolic pathways. In the metabolic pathways table, a main column corresponding to the mapping is added, comprising the following sub-columns:
  • Propagate: the percentage of mapped biological objects (reactions, proteins, etc ...) in this pathway
  • Nb mapped: The number of biological objects mapped in this pathway
  • p-value: the over-representation of the mapped objects in each pathway is tested using a Right tailed Fisher Exact Test. The p-values are corrected to account for the multiple tests performed for all pathways. Bonferroni and Benjamini-Hochberg corrected p-values are presented in the grid. *** indicates a p-value < 0.0001; **indicates a p-value < 0.001; *indicates a p-value < 0.05.




Pathway enrichment from a metabolite mapping

Reaction (respectively Metabolite) mapping adds also mapping columns into Metabolite (respectively Reaction) grid with coverage values. Values in additional colums are also reported in the Pathway, Metabolite and Reaction grids if the option "Without condition values" is deselected in the Propagate menu in the Mapping form. Three options are possible:
  • with conditions (min value): the minimum value among all the columns is reported
  • with conditions (max value): the maximum value among all the columns is reported
  • with conditions (average value): the average value among all the columns is reported

MeSH mining : Extract metabolic information from literature

MeSH mining from metabolite

This feature allows to get MeSH term associated with fingerprint. MeSH (Medical Subject Headings) is the NLM controlled vocabulary thesaurus used for indexing articles for PubMed. MeSH mining retrieve all MeSH terms significantly associated for each metabolite of fingerprint. The function relies on Metab2MeSH service.

Launch

To launch MeSH mining from Metabolite you need to select "MeSH mining" and "MeSH mining from Metabolite" in header "Omics" menu.
After it you have do do a mapping of metabolites to select your fingerprint.

Then you can launch MeSH mining from metabolite.



Launch MeSH mining from Metabolites


The results will appear in the Jobs grids when they will be finished. To get the result click on result button in job tab in side component.



Get MeSH mining from Metabolite results


In MetExplore « metabolites » panel you can see the metabolites used as fingerprint and a link on corresponding PubChem page. The results of MeSH mining are displayed in a dedicated heat map described below in the documentation (click here to access this part of the documentation).



Get MeSH mining from Metabolite results

Filter results

Metabolites may be associated to a large number of MeSH terms. Nevertheless, MeSH being a thesaurus, it is possible to use the underlying hierarchical structure to select a sub-part of the MeSH term thesaurus. For instance, one can wish to focus on Hepatic insufficiency related terms as show in the following figure.



MeSH filtering priciple To do that select one or multiple MeSH term with the dropdown list.



Select MeSH term as filter

Use MeSH pair Metrics as filter

MeSH filter may be hard to define since it requires to have knowledge on the content of the thesaurus. Here we present another way to filter Metab2MeSH. The overall idea is to detect which MeSH terms are significantly associated in the literature and to use that to guide the filtering. The approach uses an article-based metric which provides a MeSH term pairs that co-occur more often than expected by chance. Calculation of odds ratios is describe in this article: Smalheiser, Neil R., and Gary Bonifield. “Two Similarity Metrics for Medical Subject Headings (MeSH): An Aid to Biomedical Text Mining and Author Name Disambiguation.” Journal of Biomedical Discovery and Collaboration 7 (2016): e1. PMC. Web. 13 Feb. 2018. To use it you have to select MeSH pair metrics in the dropdown list. If the list is empty it's necessary to launch MeSH pair metrics job.



Open MeSH pair metrics form


To do that click on "Define new filters by similarity to fill previous list" button. The newly opened window allows to launch MeSH pair metrics job. Select a MeSH term in the dropdown list, click "Launch" button and the results will appear in the “Jobs” grid when computation will be finished. To get the result click on the folder icon in job tab in side component and the dropdown to select MeSH pair metrics job will be filled.



Launch MeSH pair metric and get results


Once selecting MeSH pair metrics a sorted table appear with the MeSH term that co-occurred with your interest MeSH term. Next you can select one or multiple MeSH term and launch MeSH mining from metabolite.



MeSH filtering from MeSH pair metric

Heat map visualization

To display results, MetExplore provides a heat map with metabolites in x-axis and MeSH term in y-axis. This representation provides insights on:

  • the fact that metabolites are significantly associated with MeSH terms
  • the number of metabolites of the fingerprint associated with a MeSH term and its ratio with all article containing the MeSH term
The heat map is interactive allowing:
  • sorting the heat map
  • clustering metabolites and MeSH term and reduce the heat map by cluster
  • image export
  • have information on MeSH terms and metabolites
  • find co-currency between selected nodes




Open MeSH pair metrics form


Sorting the heat map

To sort the heat map, click on right top sort button and click "Sort by metabolites". Sorting can be achieved based on:
  • number of metabolites of the fingerprint associated with a MeSH term
  • ratio between the number of metabolites of the fingerprint associated with a MeSH term and all article containing the MeSH term
  • fact that metabolites are significantly associated with MeSH terms.




Sort heat map rows


Clustering

Hierarchical clustering of metabolites and MeSH are displayed on the side of the heat map using dendrogram.
The clustering is calculated with the ward algorithm and euclidian distance.
The dendrogram allows to highlight a part of heat map or reduce to this part respectively by clicking on branch or doing a long click on branch.



Clusters highlighting


Export heat map

To export the heat map, click on right top photo button and click "Export image".



Metabolites and MeSH information

Some elements of information are available by clicking on:
  • heat map nodes (rectangles)
    • links to PubMed with a request to have all articles which are annotated with the corresponding metabolite and the MeSH terms
    • link to MeSH web page of the MeSH term
    • link to MeSH web page of the metabolite.
  • metabolites
    • link on PubMed to see articles where the metabolite is mentioned
    • link on MeSH web page of the metabolite
    • a chart with the number of publications by year where the metabolite is annotated, with the possibility to access to all article for one year.
  • MeSH terms
    • link on PubMed to see articles where the MeSH term is annotated
    • link on MeSH web page of the MeSH term
    • a chart with the number of publications by year where the MeSH term is annotated, with the possibility to access to all article for one year.




Heat map information


Find co-currency between selected nodes

This feature allows to launch PubMed request to retrieve all articles annotated by metabolites and/or MeSH terms.



Find co-currency between L-glutamic acid AND L-methionine AND L-phenylalanine AND Hepatic Encephalopathy


MeSH mining from MeSH

Get metabolites of the network associated with a MeSH term.

Launch

To launch MeSH mining from MeSH you need to select "MeSH mining" and "MeSH mining from MeSH" in header "Omics" menu.
After it you have to select a MeSH term in the dropdown list.

Then you can launch MeSH mining from MeSH.



MeSH mining from MeSH


The results will appear in the « Jobs » grid when they will be finished. To get the result, click on folder icon in the “Result” column of the « Jobs » grid.



Get MeSH mining from MeSH results


This analysis returns a p-value corresponding to Fisher exact test to show metabolites significantly cited with a MeSH, a Benjamini-Hochberg correction of the p-value and a link on PubMed with a request to have all articles which are annotated with the metabolite found and the chosen MeSH term.



Get MeSH mining from MeSH results

MetaboRank : a network based recommendation system to interpret and enrich metabolomics results

Find insightful metabolites that are well connected to your metabolites of interest obtained from experimental data. This tool uses an extended version of Recon2, the highly curated reconstruction of the human metabolic network, which contains only relevant connections selected based on biochemical criterions. It internally uses a recommendation algorithm inspired by what can be found on social networks to suggest you new users based on their network proximity.

This tool executes personalized Page Rank and Chei Rank algorithms:
C. Frainay, et al. "Metabolites you might be interested in: a network based recommendation system to interpret and enrich metabolomics results", Bioinformatics (2017), 31(20):3383–3386. | Submitted |

Use MetaboRank

Launch

To launch MetaboRank you need to select "MetaboRank" in header "Omics" menu.
After it you have do do a mapping of metabolites to select your fingerprint.

Then you can launch MetaboRank. This analysis return Page rank and Chei rank of each metabolite of the network.



Launch MetaboRank The results will appear in the Jobs grids when they will be finished. To get the result click on result button in job tab in side component.



Get MetaboRank results

Use fold change

MetaboRank allows to use fold change as weight in the random walk algorithm.
To do that you have to do a mapping of metabolites with a fold change condition to map fold change on our fingerprint and to select the corresponding condition in the MetaboRank form.

So metabolites near than metabolite of the fingerprint with big fold change could best recommendation.

Interpreting results

Page Rank & Chei rank

This analysis return Page rank and Chei rank of each metabolite of the network.
In the metabolite grid, a new column will be displayed with a three sub columns:

  • "Fingerprint" to show the metabolite of the fingerprint
  • "MetaboRank Out" for the score of the Page rank algorithm
  • "MetaboRank In" for the score of the Chei rank algorithm




MetaboRank results in grid

Scatter plot

In order to have intuitive representation of MetaboRank, MetExplore provides an interactive scatter plot.



MetaboRank results in scatter plot

It allows to :

  • Quickly see suggestions. So metabolites with high "MetaboRank In" and high "MetaboRank Out"are suggested
  • Select node or multiple nodes
  • Map selected node in MetExplore metabolites table in order to add good suggestion to relaunch the analysis and extend the fingerprint
  • Set the axis scale in log to spread out point
  • Filter node in "Fingerprint" or "Not in fingerprint" by clicking in the bottom caption
  • Export images in the right top