Visualization and Phospholipid Identification (VaLID): online integrated search engine capable of identifying and visualizing glycerophospholipids with given mass

Motivation: Establishing phospholipid identities in large lipidomic datasets is a labour-intensive process. Where genomics and proteomics capitalize on sequence-based signatures, glycerophospholipids lack easily definable molecular fingerprints. Carbon chain length, degree of unsaturation, linkage, and polar head group identity must be calculated from mass to charge (m/z) ratios under defined mass spectrometry (MS) conditions. Given increasing MS sensitivity, many m/z values are not represented in existing prediction engines. To address this need, Visualization and Phospholipid Identification is a web-based application that returns all theoretically possible phospholipids for any m/z value and MS condition. Visualization algorithms produce multiple chemical structure files for each species. Curated lipids detected by the Canadian Institutes of Health Research Training Program in Neurodegenerative Lipidomics are provided as high-resolution structures. Availability: VaLID is available through the Canadian Institutes of Health Research Training Program in Neurodegenerative Lipidomics resources web site at https://www.med.uottawa.ca/lipidomics/resources.html. Contacts: lipawrd@uottawa.ca Supplementary Information: Supplementary data are available at Bioinformatics online.


I. What is VaLID?
"Visualization and Phospholipid Identification" (VaLID) is a web-based application linking a convenient search engine, a phospholipid database, and multiple visualization features for identification and dissemination of large-scale lipidomic datasets. Given the rapidly evolving nature of the lipidomic field, many mass to charge (m/z) values are not yet represented in other open-access web-based engines. To address this need, VaLID returns all theoretically possible species based on m/z and user-defined MS conditions in addition to highlighting species identified by members of the Canadian Institutes of Health Research (CIHR) Training Program in Neurodegenerative Lipidomics (CTPNL). The intent is to complement existing curated databases by enabling users to predict the identity of 'new' species detected in their particular mass spectrometry (MS) datasets and thus proceed more rapidly to validation. While useful for lipid discovery, the user is cautioned that VaLID includes lipids (and isomeric bond configurations) that may not be biologically relevant. Investigators are encouraged to mine these lists for species most relevant to their specific biological system for subsequent validation. To assist in decision-making, a "best guess" feature is available whereby lipids listed in blue are predicted to be the most likely based on the prevalence of constituent fatty acid chains in mammalian cells (Miyazaki and Ntambi, 2008). Every theoretical conformation (in cis configuration) for each species can be viewed in 2D and 3D. In this version, curated species detected in brain tissue by members of the Canadian Institutes of Health Research (CIHR) Training Program in Neurodegenerative Lipidomics (CTPNL) can also be downloaded in multiple high-resolution representations for further visualization and model production.

Running VaLID
2.1. You will be asked to accept the security certificate for our lipid database. This message will not appear again once you have accepted the certificate.
In some browsers, VaLID will not start until you click once in the centre of the browser window to initiate Java. Once properly loaded, this window will appear: Note: The glycerophospholipid database contains phospholipids with fatty chains from 0 to 30 carbons and up to 6 cis unsaturations. The differential combinations of these individual fatty acyl chains at the sn-1 and sn-2 positions with ester, alkyl ether, or vinyl ether linkages lead to the formation of the majority of biologically relevant lipids. 3.4. When all of the search options are selected based on your MS conditions, click on the "Search" button or press the "Enter" key on the keyboard. Note: The program currently does not support parallel searches of multiple lipid species in different tabs of the same web browser. 3.5. Initial searches may take longer than a minute, depending upon the connection speed, and search options selected. Why? When a new search is started the program loads all the lipids with the mass you are searching for into memory, as well as lipids with mass ratios 25 m/z above and below the target mass. This cached buffer of lipids with 50 m/z of your target range allows for subsequent searches close to the original search to occur much faster than if the program reloaded the database each time. 3.6. The possible lipids based on the search criteria will appear in two panels, (i) "Possible Lipids Include" and (ii) "Possible Isomeric Lipids Include." 3.7. The left panel (Possible Lipids Include) shows lipid predictions with different chains at sn-1 and sn-2 positions of the glycerol backbone while the right panel (Possible Isomeric Lipids Include) shows corresponding isomers with the same chains but at different positions on the glycerol backbone. Lipid location in these lists is not sorted by biological significance but grouped according to lipid subclass (polar head group) and ascending (i) m/z, (ii) number of carbons (sn-1 chain), (iii) degree of unsaturation (sn-1 chain), (iv) number of carbons (sn-2 chain), and (v) degree of unsaturation (sn-2 chain).
3.9. Each lipid entry is selectable (click once to highlight a given species). Name and m/z value can be copied by pressing "CTRL + C" and pasted by "CTRL + V". 4.2.2. The "Best Prediction" button displays only lipid species that are predicted to be the best guess. This function works only with lipid entries in blue font. Note: Lipids in VaLID's "Predicted to be Common" database were identified based on the relative abundance of certain fatty acid chains in mammalian cells (Miyazaki and Ntambi, 2008). If both of the fatty acid chains at the sn-1 and sn-2 positions or if a single chain in a lysophospholipid are considered common, the lipid is included in the Predicted Common database and appears in blue.

Displaying glycerophospholipid structures
4.2.3. The "Structural representations" button displays lipids that are part of our CTPNL curated database of neural lipids. This button displays a known structural representation present in mammalian membranes as a (i) 2D skeletal model, (ii) 3D "Ball and stick" model, (iii) "Space filling" model and (iv) our own rendered "VaLID view" model.

Downloading glycerophospholipid structures
5.1. To save a lipid structure from the "Display All" or "Best Prediction" windows, double-click on a structure of interest in the pop-up window displaying all possible configurations. A new window will open with the selected lipid structure.
5.1.1.2. As a chemical structure information file that is compatible with ChemDraw and Marvin, go to the "File" menu and choose "Save Selection." A save dialogue box will appear, allowing you to save the file on your hard drive. You can save the file as one of many file formats, including ChemAxon Marvin (.mrv), MDL Mol file (.mol) and CDX file (.cdx, suitable for ChemDraw) formats.
6. Exiting the application 6.1. To close the application, navigate away from the page. 6.2. To free up all resources the application has used, quit/close your browser. By closing your browser, you close Java and any system resources used by the application. Typically, these tools are used in the animation of dynamic systems such as fire, smoke and water. nParticles are easily converted into smooth and organic polygonal meshes. By directing these particles to the x, y, and z coordinates originally generated in ChemDraw 3D and imported in Maya as points in space, a mesh that recapitulates the original molecular structures in an abstracted, organic, form. Here, VaLID view models are available for download as rigid polygon models however in Maya, we have fitted each species between each atom with a rig of movable joints, a process typically used by graphic artists to animate human or animal characters. By creating a skeleton embedded within the mesh model, we have enabled each of these structures to move in a 'biological' or dynamic way. Joint attributes are modified to restrict or limit movement (i.e., in the case of a double bond) and can be reassembled into membrane form.