As research employing lipidomics is rapidly increasing there is a great need for bioinformatic tools that capture and utilize the complexity generated from lipidomic analyzes. Currently, the diversity and complexity in the lipidome is often under-described by averaging lipids up to the class-based descriptors, losing valuable information about biological function and interactions with other lipids, proteins and/or metabolites. While metabolomics methods and tools have improved over the past few years, computational developments are at the center of the gap in deriving insights from lipidomics data.

Lipids play crucial roles in a plethora of physiological functions ranging from energy storage, cellular compartmentalization, regulation of protein function to signaling. In order to unravel lipid function it is of utmost importance to identify and quantify single lipid molecular species in complex biological mixtures. Lipidomics aims to describe the whole variety of lipid species and to provide the knowledge on their diversity, distributions, and concentrations suitable for further systems-wide data integration strategies. Despite successful development of analytical solutions in mass spectrometry (MS) based lipidomics, reliable high-throughput identification of lipids form (LC)MS/MS datasets remains one of the main bottlenecks. To address this challenge, we developed several software for covering identification of lipids (LipidHunter), modified lipids (LPPtiger), and lipid annotation converter (LipidLynxX). In my presentation I will cover main principles of accurate lipid identification strategies and computational solution towards their high-throughput applications.

SwissLipids (www.swisslipids.org) is a freely available, expert-curated knowledgebase of lipids aimed at helping integrate lipidomics data with lipid biology. Lipid metabolism is described using biochemical reactions from the Rhea knowledgebase (www.rhea-db.org) and enzymes from the UniProt knowledgebase (www.uniprot.org). SwissLipids currently comprises more than 590,000 lipid structures from over 550 lipid classes and over 5,000 enzyme-reaction pairs. SwissLipids features a hierarchical classification that links lipid analytes to structures and enzymes like this: [PC(38:4) -> PC(18:0_20:4) -> PC(18:0/20:4) -> PC(18:0/20:4(5Z,8Z,11Z,14Z)) -> PLA2G4A]. This classification is compatible with the nomenclature recommendations of the Lipidomics Standards Initiative and the LIPID MAPS classification. All data is freely available and accessible to search and browse via our API and website. Our future plans include leveraging UniProtKB as the basis to build an enhanced SwissLipids v2.0 – eventually covering all possible products of lipid metabolizing enzymes and taxa from UniProt.

Lipidomics analyses enable the identification and the quantification of hundreds of lipids creating rich datasets. A challenge that arises with lipidomics is the availability of tools to assist with data analysis to further the biological interpretation of complex lipids (e.g., phospholipids, glycerolipids, and sphingolipids). While enrichment analysis is a common starting point in sequence-based omics studies, such tools were lacking for lipidomics. We have developed a lipidomics enrichment tool entitled Lipid Mini-On that is comprised of an R package and a Shiny user interface. Lipid Mini-On enables the generation of structural ontology terms by mining the lipid common names. Enrichment statistics are then performed on these terms. Lipidomics data interpretation is facilitated by visualization functions implemented in the tool suite. Lipid Mini-on has been used for diverse applications including environmental and biomedical sciences. We will demonstrate its utilization to interpret spatiotemporal trends occurring in the developing lung. Other tools available to perform pathway and enrichment analysis will be discussed. We anticipate that the new tools developed by us and others will ease the data analysis journey and further the biological knowledge gained from lipidomics analyses.

Hepatic stellate cells (HSCs) are professional lipid-storing cells and are unique in their property to store most of the retinol (vitamin A) as retinyl esters in large-sized lipid droplets. HSC activation is a critical step in the development of chronic liver disease. After liver injury, quiescent HSCs can transdifferentiate into activated cells with a myofibroblastic phenotype, a process which involves dramatic rearrangements of the lipid composition. Identification of metabolic pathways and key nodes that are involved in these changes will be crucial for the development of drugs that interfere with HSC activation and prevention of liver disease. A major challenge in these types of lipidomic analyses is the handling of the large amounts of data and the translation of results to interpret the involvement of lipids in biological systems. We built a new lipid ontology (LION) that associates >50,000 lipid species to biophysical, chemical, and cell biological features. By making use of enrichment algorithms, we used LION to develop a web-based interface (LION/web, www.lipidontology.com) that is freely available to the scientific community.

Nonalcoholic fatty liver disease (NAFLD) is a progressive liver disease that is strongly associated with type 2 diabetes. Accurate, non-invasive diagnostic tests to deliniate the different stages: degree of steatosis, grade of nonalcoholic steatohepatitis (NASH) and stage fibrosis represent an unmet medical need. Here we report underlying associations between lipidomic profiles, metabolic profiles and clinical outcomes, including downstream identification of potential biomarkers for various stages of the disease. Data were interrogated on patients representing the full spectrum of NAFLD/NASH, derived from the EPoS European NAFLD Registry (n = 627). By using patient-matched liver transcriptomics and serum metabolomics/lipidomics data (n = 223) from the same cohort, we also conducted genome-scale metabolic modeling to dissect hepatic metabolism across in NAFLD. We found that steatosis grade was strongly associated with (1) an increase of triglycerides with low carbon number and double bond count as well as (2) a decrease of specific phospholipids, including lysophosphatidylcholines. Our findings suggest that dysregulation of lipid metabolism in progressive stages of NAFLD is reflected in circulation. A detailed network-based picture emerges between lipids, polar metabolites and clinical variables. Lipidomic markers may provide an alternative method of NAFLD patient classification and risk stratification to guide therapy.

With the increased power to analyze the lipidome, the availability of bioinformatics approaches becomes urgent. Similar to metabolomics, the interoperability between experimental data and pathway databases helps our systems biology understanding. Shared interoperability challenges include that the measured lipid structure of the may not always be known in detail, and that we need to handle lipid classes. Mirroring, the biological knowledge captured in a pathway database may also refer to lipid classes, due to limited analytical power in the past. Complications unique to lipidomics include the aggregation of lipids (e.g. lipid rafts) which require new interoperability approaches. This presentation will give an overview hpw WikiPathways, BridgeDb, and Scholia integrate information from multiple biological resources. WikiPathways captures the biological pathways, BridgeDb provides us with links out to third-party databases, and Wikidata and Scholia link lipids to primary literature. We will discuss the issues of chemical identity (from a cheminformatics perspective), compound classes (e.g. the LIPID MAPS ontology in Wikidata), lipid complex representation, and how these aspects affect system biology data analysis. Where possible, we will demonstrate this using the pathways on the WikiPathways Lipids Pathways Portal (http://lipids.wikipathways.org/) which includes pathways provided by the LIPID MAPS project.