MetagenoNets Tutorials

Welcome to Tutorials Section for MetagenoNets

Here we have attempted to provide you with precise guidelines for using MetagenoNets. You can use the left menu to fast-scroll across the content. .

Why MetagenoNets?

Motivation

Microbial association networks (often termed as Co-occurence networks) are very frequently used in Microbiome research to understand community dynamics (or interactions). Inferring any biological network and obtaining relevant biological insights however requires a lengthy workflow of data management, choice of appropriate methods, statistical computations, information analysis, followed by an entirely different pipeline for suitably visualizing, reporting and comparing the interactions. This complexity of workflow is further increased with the added dimension of multi-class and multi-group 'Meta Data' often associated with Microbiome studies, thereby giving rise to a need for not only classified networks, but also integrated and bipartite networks. The plethora of correlation analysis methods ranging from simplistic Pearson, Spearman, Kendall etc. to statistically intensive methods like ReBoot, CCREPE, SPARCC and NAMAP further increase the amount of effort required to perform correlation based microbiome interaction studies.

To date, there is a high dependency on stand-alone generic softwares, plugins, locally installed programs (with a series of dependencies), knowledge of scripting/ programming languages for accomplishing the task of a meaningful microbiome network analysis. Limited number of web applications available in this space are either too specialized for other research areas or offer minimal functionalities. A comprehensive tool (especially a web-server) that may circumvent most of these bottlenecks, would be highly useful to the researchers at large.

Features in MetagenoNets

MetaNets as a Solution

We present MetagenoNets web-server as an effort towards easing the process of inferring and analyzing correlation driven microbial association networks. Following features of MetagenoNets are expected to be of significant value addition to the space of microbiome network analysis:

(a) Accepts all types of small to large microbial feature tables along with multi-level meta-data. Provision for secondary feature tables allows deeper insights for an integrated analysis (as described later in this tutorial).
(b) Offers frequently used data normalization startegies (TSS, CSS, Quartile) and transformation strategies (DeSeq2, TMM, CLR)
(c) Provision for feature reduction through prevalence and occurence based filters
(d) Comprehensive offering of major state of art correlation driven network inference methods (i) SPARCC (ii) ReBoot (iii) CCREPE (iv) NAMAP, alongwith bootstrapped and classical versions of approaches based on Pearson and Spearman coefficient.
(e) Intelligent categorization of Meta-data into categorical and continous data types, thereby allowing the end-user to automatically correlate continous meta-data with primary feature set through integrated and bi-partite networks (as described later).
(f) Three types of network generation options (i) class or group respective individual networks (ii) meta-data and/or secondary data integrated networks (iii) bi-partite networks, thereby enabling a comprehensive use of abundance datasets as well as associated meta-data
(g) Interactive visualizations for networks and network properties
(h) Set based comparisons of categorical meta-data driven network groups through interactive Venn Diagrams and network graphs of the intersections
(j) Easy to use user inter-face and modern web-design approach for seamless experience at the front-end

MetagenoNets facilitates end-users to concomitantly infer, statistically analyse, and compare correlation driven microbial association networks, and in the process generate a plethora of intuitive self-explanatory visual outputs in an automated fashion.

Input Data and Meta Data Formats

Input Data for MetagenoNets

MetagenoNets accepts two types of input data, (i) Primary Input Data (ii) Secondary Input Data

Primary input data is essentially a multivariate abundance table representing abundances of various operational taxonomic units (OTUs) obtained from various de-novo or reference based taxonomic classifiers. It should be a tab-separated file, wherein first row pertains to the samples in the data, and first column pertains to the taxa (or classified microbes) in the data. It is mandatory to provide a Primary Input Data to MetagenoNets.

Secondary input data is an optional input data type, that is also a multivariate abundance table. However, this input table may contain any feature matrix (like Pathway abundances, Metabolite abundances, Continous Data of any kind) for the samples provided in the primary input data. This feature set is used for finding relevant set of co-variates against the primary feature set. The co-variates are thereafter visualized in 'integrated' as well as 'bi-partite' networks, as preferred by the end-user.

Primary Input Data

Download Sample

Secondary Input Data

Download Sample

Meta Data for MetagenoNets

Meta data for MetagenoNets is a tab-separated file containing multiple columns of sample information. First row contains the names of various meta data classes, while first column contains the names of the samples (as present in the input data).

Each column of meta data file, representing various classes of environments, contains the names of various sub-classes in each environment.

It is pertinent to note that MetagenoNets automatically infers the categorical and continous metadata types in the supplied meta-data file. The categorical data is offered to the end-user for creating various class-specific of networks, while continous data is offered to find co-variates in the primary and secondary input data feature set. The co-variates are visualized in the integrated as well as bi-partite networks.

Simple Meta Data

Download Sample

Integrated Meta Data

Download Sample

Summarizing the Typical Format

Typical Input and Meta Data Format

JOB-ID Management

All modules of MetagenoNets follow a unique 10-character alphanumeric JOB ID based job handling and perform all analyses without seeking any personal information. Coupled to that, each JOB ID is subsequently tagged to a personal Dashboard, which packages and presents all visualizations and results. All the dashboards, corresponding to user-specific job IDs, can be downloaded as unique 'dash files' which can be retained and shared by the end-users for future re-explorations. The 'dash files' can be uploaded to the specifically developed 'Recreator' module of MetagenoNets, which recreates an entire dashboard using the unique dash file.

One can access the Results by pasting the unique JOB ID into the JOB Status input fields provided on top of each module of MetagenoNets in both the Themes (following are the snapshots):

For Deep Theme

JOB Staus Checking Option

For Lite Theme

JOB Status Checking Option

One can also access his/ her JOB History on MetagenoNets using the floating button on the right side of the execution section of each module in the Deep Theme mode and using the navigation bars in the Lite Theme (following is a snapshot):

For Deep Theme

JOB History Access

For Lite Theme

JOB History Access

Clicking the JOB History button shows a pop-up notification on the bottom right of the screen, displaying the number of jobs in your current history, and a button to access your JOB History page (following is a snapshot):

History Exists

Notification when jobs exist in the history

Nothing in History

Notification when there is nothing in the job history

If a history exists, you'll be able to access the same using the Access JOB IDs button provided in the notification:

Job History Page

A Snapshot of JOB History Page

Output Data and Results

Each module of MetagenoNets generated two broad category of outputs or results:
(a). Ensemble of Textual Results
(b). Ensemble of Visual, Interactive Results in personalized Dashboards

Textual Results

Each module of MetagenoNets generates a comprehensive ensemble of textual matrices or files corresponding to module specific functional inference and analytical workflows.
For example, the textual esemble of Global Mapper results typically comprises of the following kind of files:

1. Raw Function Abundance Profiles: Unfiltered Function abundance profiles for all samples of the environment are provided as L3.profile.txt, L2.profile.txt and L1.profile.txt. L3,L2 and L1 represent the specific levels of functional hierarchy.
2. Function Abundance Profiles at PEC Thresholds: The functional abundance profiles are inferred for all levels of hierarchy at PEC 50 - 90 thresholds (what is PEC?) and provided in well stuctured directories.
3. Bacterial Contributions: Functional contributions by various microbes in the environment in each class of samples is appropriately segregated and presented in well structured directories. Contributions at various PEC thresholds are also provided in separate directory structure.
4. Analysis Directory: A directory containing various analysis outputs like Core Functions, Top Functions, Networks, Differentiating Functions, Ordination is also provided. A separate sub-directory exists for each kind of analysis and contains multiple files pertaining to that analysis.
and more..

Download Sample Textual Results for Global Mapper

Download Sample

A similar kind of structure and module specific ensemble of textual results are created for ISFA and Local Mapper as well.

Visual and Graphical Results (DASHBOARDS)

Each job execution on every module of MetagenoNets is tagged to a unique JOB ID which in turn is linked to a full-fledged dashboard of results and analyses. A dashboard represents an ensemble of carefully compiled personalized web-page that showcases various graphs, interactive, visualizations, tables and downloadable files specific to the module it corresponds to.

Each module has its own style and content of Dashboard, and each user gets a personal, independent dashboard for each job he/ she executes on any module of MetagenoNets.

Following are direct links and snapshots of sample dashboards specific to Global Mapper, Feature Analyzer and Local Mapper: