For either database, you can further interrogate "bulk" structures returned, to generate a list of potential structures from LMSD. COMP_DB search results also contain links, where appropriate, to lists of structures computationally-generated from all chain lengths.
COMP_DB is a virtual database composed of major classes of lipid species, generated from a list of commonly occuring acyl/alkyl chains. These "bulk" lipid species indicate the number of carbons and number of double bond equivalents, but not chain positions or double bond regiochemistry and geometry.
COMP_DB will generally return a higher number of lipids, however many of those may not be relevant to mammalian samples. If users prefer to search on a list that is more applicable to plasma or cells, then LMSD is recommended.
The LIPID MAPS Structure Database (LMSD) is a relational database encompassing structures and annotations of biologically relevant lipids. One can use this interface to search for "bulk", or "chain composition", structures — which provide information on number of carbons and double bond equivalents, but not regiochemistry or geometry — found in LMSD. While only "bulk" lipid species are shown in the initial search results, links from these "bulk" lipid species to corresponding fully annotated discrete lipids found in LMSD are provided.
As of 02/28/2020, LMSD contains 43665 unique lipid structures, making it the largest public lipid-only database in the world. Please note, however, that only lipids in categories Fatty Acyls [FA], Glycerolipids [GL], Glycerophospholipids [GP], and Sphingolipids [SP] are included in this search. Lipids in categories Sterol Lipids [ST], Prenol Lipids [PR], Saccharolipids [SL], and Polyketides [PK] are not amenable to representation by a "bulk" abbreviation.
Please see Shorthand notation for lipid structures derived from mass spectrometry. J Lipid Res. 2013 Jun;54(6):1523-30. doi: 10.1194/jlr.M033506 to learn more about how these "bulk" lipids are annotated.