DATA ANALYSIS, INTEGRATION & STEWARDSHIP

Recently, we have seen an explosion of -omics data. Many analysis pipelines for genomics, transcriptomics, proteomics and metabolomics are available, but to make integrated use of those data and to translate them into actionable knowledge (e.g. new bioengineering strategies, new diagnostic routines, new drug targets) is a major bottleneck.

This is particularly true for researchers, who do not have access to state-of-the-art bioinformatics expertise, tools or analysis environments. The X-omics infrastructure will enable efficient data integration for data generated within X-omics and external datasets. X-omics adopts the FAIR principles and takes them a step further by helping researchers to: make data FAIR at the source, facilitate reuse of their data and make their data analysis workflows transparent and reproducible.

To facilitate analysis, integration & stewardship of X-omics data, X-omics will start developing:

An integrated strategy for experimental design that will enable the analysis of complex samples with multiple omics technologies.

A data infrastructure based on FAIR principles and design methods to combine different data sources from analytical omics-platforms and other related data to obtain a comprehensive view of complex systems.

Development of more sensitive enrichment technologies to improve depth of analysis of post-translational protein modifications.

Data analysis, integration & stewardship services

The main area of expertise of the data facilities, that are part of the X-omics research infrastructure, consists of integration of X-omics data, FAIRification of -omics data, reproducible data analysis and workflow management. The services are described below.

Upstream data analysis

Diagnostics quality variant calling
Long read analysis and optical mapping
Differential expression analysis
Quantitative, LC-MS & phospho-proteomics analysis
Quantitative Metabolomics analysis
UntargetedMetabolomicsanalysis
Methylation analysis
Microbiomics analysis
Lipidomics analysis

Downstream data analysis

Somatic variant analysis and structural variations
RNA Isoform & splicing analysis
Pathway analysis for RNA-Seq, Methylation and Untargeted Metabolomics
Next-generation metabolic screening
Metabolite set enrichment analysis
Genotype linkage to phenotype and rare diseases
FAIRification, Data Ontology support and Text mining

Multi-omics data analysis & integration

Proteogenomics combining Proteomics with DNA, RNA or Long reads
Pathway and gene network integration
Statistical integration of multi-omics data

All services mentioned above are custom services.
Please contact our helpdesk for a quote with turnaround times and prices based on your research question.

Helpdesk

FAIR Data Cube

The FAIR Data Cube (FDCube) is an important infrastructure component in the Netherlands X-omics initiative. It is a set of tools and services that helps researchers in distinct stages of the Research Data Life Cycle with:

Creating rich and machine readable metadata for single and multi-omics datasets;
Making -omics data findable and accessible for reuse;
Facilitating analysis and integration of -omics data from different sources.

Creation of multi-omics data and metadata

In collaboration with the FAIR Genomes project and United for Metabolic Diseases (UMD), we develop metadata schemas for different omics data types. We use the Investigation-Study-Assay (ISA) metadata framework to capture experimental metadata and share example workflows for metadata creation.

Querying multi-omics studies

Researchers can find studies registered in a FAIR Data Point (FDP). Semantic searches on samples, phenotypes, omics measurements and features are supported by an RDF-based knowledge graph.

FAIR data stewardship and tool development in X-omics

Github: Fair Data Cube

Multi-omics analysis

Over the past years the X-omics consortium data analysis core has gained a lot of experience working with multi-omics datasets and integration. We have identified several tools and strategies per omics data analysis approach and study design which we have formulated in the following analysis matrix:

Eatris+ Github

When starting a new multi-omics data analysis we have defined several analysis flows that lead to the best project progression and results. Here we show a typical data analysis flow for the integration of metabolomics and epigenomics data, for example.

X-omics Github

Another example of a novel approach for multi-omics data integration and feature extraction is the use of autoencoders. Autoencoding is well established machine learning method that aims to recreate the same data from an input dataset, while it iteratively reducing the internal dataset dimensionality and complexity, captured in an increasingly smaller set of neural network layers. This mechanism reduces dataset noise and removes dimensions that do not contribute to variability. After the minimal level of Autoencoders is established, latent representations and variables become clear, which can lead to new biomarkers or biological relationships in the data that were not visible in the single omics datasets.