A website for metabolomics data analysis and mining Sara Cardoso 1, - PowerPoint PPT Presentation

WebSpecmine : A website for metabolomics data analysis and mining Sara Cardoso 1, *, Telma Afonso 1 , Marcelo Maraschin 2 , and Miguel Rocha 1,* 1 CEB - Centre Biological Engineering, University of Minho, Campus of Gualtar, Braga, Portugal; 2 Plant Morphogenesis and Biochemistry Laboratory , Federal University of Santa Catarina, Florianpolis, SC, Brazil. * Corresponding authors: saracardoso501@gmail.com; mrocha@di.uminho.pt 1

WebSpecmine : A website for metabolomics data analysis and mining 2

Introduction Metabolomics ✓ Analysingmetabolomics data correctly and efficiently is nowadays very important in biological and biomedical research. However Most people that want to perform this analysismay not have the programming skills needed A website to perform metabolomicsdata analysisis an importantasset 4

Introduction Some of the Existing Websites Univariate Multivariate Other Covered Techniques Analysis Analysis Features LC/GC-MS Raw Spectra PCA; Clustering; Correlation T-Test; ANOVA; Machine Learning Analysis; LC/GC-MS Peak Lists Fold Change (only PLS-DA); Metabolite NMR Peak Lists Feature Selection Identification (only Metabolites' Concentrations (only Random for MS); Pathway (Quantitative Data) Forests and SVM) Analysis PCA; Clustering; Metabolite T-Test; ANOVA; LC/GC-MS Raw Spectra Machine Learning Identification; Non-Parametric (only LDA, PLS-DA Pathway Analysis; Tests; and Random User Account Forests) 5

Introduction What is missing in the existing websites? A wide variety of techniques and SpectralData (Raman, UV-Vis and IR) is missing data formats supported A wide variety of pre-processing Mostly just normalization, scaling, missing values methods treatment A wide variety of analysis There should be more model options for machine methods learning, for example Most of the time, users have to follow a strict Flexible Pipeline pipeline So that data and results can be stored and shared User Account 6

What was our main goal, then? ✓ Create an easy-to-use and freely available website that provides a wide variety of methods and data types for analysis, and ways to store and share metabolomics data and the results generated. SOLUTION: 7

WebSpecmine : overview Metabolomicsdata Supported User Account ✓ NMR ✓ Store data and results privately ✓ LC/GC-MS ✓ Share data across users ✓ Infrared, UV-Visible, and Raman Spectra ✓ ConcentrationsData(QuantitativeData) Tutorials and User Guide Data Pre-Processing Metabolomicsdata Analysis Available ✓ UnivariateStatistical Analysis ✓ Unsupervised MultivariateStatisticalAnalysis ✓ Supervised MultivariateStatistical Analysis ✓ MetaboliteIdentification ✓ Pathway Analysis 8

WebSpecmine : Supported data LC/GC-MS Raw Spectra Data Formats Peak Lists Data Formats ✓ .mzData ✓ CSV ✓ .mzXML ✓ TSV ✓ .netDF 9

WebSpecmine : Supported data NMR Raw Spectra Data Formats Peak Lists Data Formats ✓ BRUKER ✓ CSV ✓ VARIAN ✓ TSV 10

WebSpecmine : Supported data Spectral Data: Raman, IR and UV-Vis Spectra Data Formats ✓ CSV ✓ (J)DX ✓ SPC ✓ MS EXCEL (.xlsx) 11

WebSpecmine : Supported data Concentrations (Quantitative) Data ✓ CSV/TSV File: Metabolites Names or Identifiers Concentrations values of each Samples' metabolite in each sample Names 12

WebSpecmine : Supported data Metadata ✓ All types of data should have a metadata file associated ✓ CSV/TSV File: Names of the metadata classes Metadata values for each metadata class in each sample Samples' Names 13

WebSpecmine : User Account Why a User Account? ✓ Main website functionalitiesare accessible without a user account ✓ But you will have to create an account if you want to: o Save and Share data and results o Leave an analysisin 'stand-by' 14

WebSpecmine : User Account Creationof a User Account To have one, users have to send an email, asking to create an account, and an email with the credentials will be sent as soon as possible. Email: webspecmine@gmail.com 15

WebSpecmine : User Account Data Projects: What is? A project is a study, or group of studies, which contains the data and metadatafor each study, as well as reports from the results obtained Projects can be: Private Public 16

WebSpecmine : User Account Your Projects The projects stored in an account are accessible through My Projects sidebar tab 17

WebSpecmine : User Account Public Projects Everyone that accesses the website can see all public projects, at the Public Projects sidebar tab 18

WebSpecmine : User Account Public Projects Everyone that accesses the website can see all public projects, at Public Projects sidebar tab To analyse a public project, you would have to copy it to your account and analyse it from there, so that the original project is not compromised. 19

WebSpecmine : User Account Workspace: Users can leave their analysis in 'stand-by' and continue later Users can leave an analysis at any time, by saving the workspace, and continue next time 20

WebSpecmine : Select Data for Analysis For Logged In Users Select the Project, the data folder 1 from that project where the data to analyse is, and the metadata file from that project that corresponds to the data selected 21

WebSpecmine : Select Data for Analysis For Logged In Users 1 Set the options required to 2 correctly read the data and metadata files 22

WebSpecmine : Select Data for Analysis For Logged In Users 1 2 After finishing the setting of data 3 and metadataoptions, the user can submit the data for analysis 23

WebSpecmine : Select Data for Analysis For Logged Out Users ✓ The Procedure is similar, but the data files and metadata files have to be submitted, as they are not stored in the website. ✓ The data submitted will only be temporarilystored, while the analysis is in action. 24

Once the user selects the data, the data analysis pages will be accessible 25

WebSpecmine : Data Visualization The website provides a way to visualize the data 1 26

WebSpecmine : Data Visualization Data Summary 1 2 27

WebSpecmine : Data Visualization Data and Metadata Tables 1 2 3 28

WebSpecmine : Data Visualization Samples' and Variables' Statistics 1 2 3 4 29

WebSpecmine : Data Visualization Boxplots of the Variables 1 2 3 4 5 30

WebSpecmine : Data Visualization Plot for Peaks Data 1 2 3 4 5 6 31

WebSpecmine : Data Visualization Plot for Spectra 1 2 3 4 5 6 32

WebSpecmine : Pre-Processing The website provides a wide variety of pre-processing methods, that can be performed in 1 the desired order 33

WebSpecmine : Pre-Processing Methods that are availablefor all types of data 1 2 34

WebSpecmine : Pre-Processing Methods that are availablefor all types of data 1 2 35

WebSpecmine : Pre-Processing Methods that are availablefor all types of data 1 2 36

WebSpecmine : Pre-Processing Methods only for spectral data 1 2 3 37

WebSpecmine : Pre-Processing Method only for NMR Spectra 1 2 3 4 38

WebSpecmine : Pre-Processing After processing the data, a name to the new dataset 1 2 3 4 5 has to be given To perform an analysis on the new dataset, the user will have to choose it on the sidebar panel 39

WebSpecmine : Data Analysis 40

WebSpecmine : Data Analysis Univariate Analysis Example for T-Test Analysis options for a T-Test 41

WebSpecmine : Data Analysis Univariate Analysis Example for T-Test Types of results available for this type of analysis: numerical results 42

WebSpecmine : Data Analysis Univariate Analysis Example for T-Test Types of results available for this type of analysis: plot 43

WebSpecmine : Data Analysis Univariate Analysis Other Analysis There are other Univariate Analysis methods available The types of results available for each analysis is similar to those showed for T-Test 44

WebSpecmine : Data Analysis Principal Components Analysis (PCA) Analysis options for both normal and robust PCAs 45

WebSpecmine : Data Analysis Principal Components Analysis (PCA) Types of results available for this type of analysis: numerical results 46

WebSpecmine : Data Analysis Principal Components Analysis (PCA) Types of results available for this type of analysis: plot results 47

WebSpecmine : Data Analysis Principal Components Analysis (PCA) Types of results available for this type of analysis: plot results 48

WebSpecmine : Data Analysis Clustering Analysis Hierarchical Clustering Analysis options 49

WebSpecmine : Data Analysis Clustering Analysis Hierarchical Clustering Types of results available for this type of analysis 50

WebSpecmine : Data Analysis Clustering Analysis K-Means Clustering Analysis options 51

WebSpecmine : Data Analysis Clustering Analysis K-Means Clustering Types of results available for this type of analysis 52

WebSpecmine : Data Analysis Machine Learning Train Models Analysis options 53