WebSpecmine

Metabolomics and Spectral Data Analysis and Mining


WHAT'S NEW

June, 2018 - Some projects from the Metabolights database are now available to analyse. NEW
May, 2018 - NMR spectra with BRUKER or VARIAN format is now supported. See help page for more information.
May, 2018 - Pathway analysis is now available.
October, 2017 - WebSpecmine, a tool for metabolomics and spectral data analysis and mining.












My Projects


You must be logged in to see your projects.

List of Projects

No projects created.

Data Folders:

Download

Files in Metadata folder:

Download

Files in Reports folder:

Download

Project description:

Public Projects


Community projects

Project description:

You must select a project first.

View project files in:

Data Folders:

Files in Metadata folder:

Files in Reports folder:

Metabolights Projects


List of Metabolights Studies Available

Click in a row to see more detailed information on the study selected.








Run Analysis


To Start the analysis of your Metabolomic Data, choose one of the analysis boxes bellow.
Boxes in grey represent unavailable boxes.

(This occurs when the dataset data type is unsupported or the dataset has missing values (treat them on "Pre-Processing" tab)).


Univariate Analysis

- T-Test
- One-way and multifactor ANOVA
- Kruskal-Wallis and Komolgorov-Smirnov tests
- Fold Change analysis

Principal Component Analysis (PCA)

- Perform principal component analysis
- Both classical and robust approaches available


Clustering Analysis

Two types of clustering analysis available:
- Hierarchical Clustering
- K-Means Clustering

Machine Learning

- Train models with the data available.
- Predict new samples with the models trained previously or a model saved in user's account.

Feature Selection

There are two methods available for Feature Selection:
- Recursive Feature Elimination.
- Selection by Filter

Metabolite Identification

Identification of metabolites only available for datasets obtained from the following techniques:
- LC-MS technique
- NMR Peaks

Regression Analysis

Available analysis:
- Regression analysis
- Correlation analysis

Pathway Analysis

Available for:
- Metabolites identified through 'Metabolite Identification' box
- Concentrations data whose variables names are in HMDB OR KEGG codes

Run Analysis

Feature Selection



Error:
Analysis name already exists or it contains spaces, please write another one.



For Model validation:

Run Analysis

Metabolite Identification


The metabolite identification is performed using the MAIT package.
Peaks are first annotated, by using the default MAIT table for adducts in positive polarization.
Next, statistically significant features are detected, followed by the identification of biontransformations between features, as well as looking for adducts.
Finally, the metabolite identification for the significant features is performed, by using the Human Metabolome Database (HMDB), version 2009/07. The peak tolerance value is set to 0.005.

ANALYSIS OPTIONS

Construction of clusters parameters:
Filtering of reference metabolites:
There are no reference metabolites with all the features selected.
Error:
Analysis name already exists or it contains spaces, please write another one.


Run Analysis

Machine Learning


Feature only available when you have trained models.

TRAIN MODELS OPTIONS

Parameter Optimization:
Model validation:
Error:
Analysis name already exists or it contains spaces, please write another one.
PREDICT SAMPLES OPTIONS

Please note that the dataset currently being used, chosen in the tab 'Dataset being used', must be the same one used for the model training.




                    
Error:
Analysis name already exists or it contains spaces, please write another one.

Run Analysis

Univariate Analysis

T-Test

Error:

One-Way Analysis Of Variance (ANOVA)

Error:

Multi-Factor Analysis Of Variance (ANOVA)

Error:

Kruskal-Wallis Test

Error:

Kolmogorov-Smirnov Test

Error:

Fold Change Analysis

(Instead of calculating the difference of the variables on two groups, it calculates the difference of the groups on two variables)
Error:

Run Analysis

Principal Component Analysis

Normal PCA

Error:

Robust PCA

Error:

Run Analysis

Cluster Analysis

Hierarchical Clustering

Error:

K-means Clustering

Error:

Run Analysis

Regression Analysis

Linear Regresssion Analysis

Error:

Correlation Analysis


Please note the larger the dataset the more time it takes to perform the analysis.
Error:

Run Analysis

Pathway Analysis


1. Choose the group of organisms where the organism wanted is:












2. Choose the organism:

3. Further options and Submit:


Error:



Pre-Processing


Missing Values

Your data has no missing values.

Data Transformation

Scaling

Correction

Smoothing interpolation

Convert to factor

Mean-Centering

First derivative

Multiplicative Scatter Correction

Data Normalization

Detect NMR Peaks


Options to align peaks after their detection: There are two methods available to perform alignment of peaks. The specmine algorithm does not allow overlapping of windows, being the size of the window equal to the step. The MetaboAnalyst method allows overlapping of windows, being the step half the size of the window. The step size for the MetaboAnalyst method has a default of 0.015 for NMR peaks and 0.125 for GC/LC-MS peaks. The bandwidth, used in this method, has the values 10, 30 and 5 for NMR, LC/MS and GC/MS peaks, respectively.
Error:

Subset Dataset

Remove data

Remove data by NAs

Low-level data fusion

Only the samples from the new data provided that have the same name as samples in the current dataset will be joined.

Note that only the formats .mzXML, .netCDF, mzData are supported. When reading the data, the peak detection will be performed.
Options for the feature (peak) detection in the chromatographic time domain:

Commonly, 30 for LC-MS spectra and 4 for GC-MS spectra.
Commonly, 30 for LC-MS spectra and 5 for GC-MS spectra.
Note that the formats supported are Processed BRUKER or Raw VARIAN. See the help page for more information on how the data folders should be formatted in each format. When reading the data, the peak detection will not be performed. To do so, you will have to go to the pre-processing page.
Options for processing the fid spectra into an intensity vs ppm spectra:
The specmine algorithm does not allow overlapping of windows, being the size of the window equal to the step:

Aggregate samples

Samples can be aggregated according to the classes of a certain metadata variable. Samples in the same class will be aggregated together.

Flat Pattern Filter

Name for the new dataset

Analysis Results

Metabolite Identification





Scores

Cluster Peaks

Top Metabolites


Reference Peaks

Matched Peaks



Cluster Peaks

No metabolites matched this cluster

Analysis Results

Feature Selection






Feature Selection Results Report (html): Download

Analysis Results

Model Training





Analysis Results

Samples Prediction

Predicted samples report (html):
Download

Analysis Results

Pathway Analysis





Loading...
Metabolites with no KEGG code ('-') were not used in the pathway analysis, as it is necessary a KEGG code to do so.

Pathways Results Table

Downloads CSV File MS EXCEL File

Saves

Metabolites Analyzed Information Table

Downloads CSV File MS EXCEL File

Saves

Analysis Results

T-Test Analysis

Analysis Results

One-way ANOVA


Analysis Results

Multifactor ANOVA


Analysis Results

Kruskal-Wallis Test Analysis

Analysis Results

Kolmogorov-Smirnov Test Analysis

Analysis Results

Fold Change Analysis


Analysis Results

Hierarchical Clustering


Analysis Results

K-means Clustering


Analysis Results

Normal PCA


Analysis Results

Robust PCA


Analysis Results

Linear Regression Analysis


Analysis Results

Correlation Analysis

Data Visualization




Dataset Visualization Report (html):
Download


The data you are exploring in this tab is the data selected in the sidebar section 'Dataset being used'.

If a metadata variable is not available to choose in the boxplots and/or spectra plots, it means that it needs to be converted to a factor (Pre-Processing page).

              












Help Page