4D Lipidomics Annotation: Deconvolving Chimeric MS/MS Spectra with Ion Mobility and CCS

Four-dimensional (4D) lipidomics represents a paradigm shift in the structural annotation of complex lipidomes. By incorporating Trapped Ion Mobility Spectrometry (TIMS) into high-resolution liquid chromatography-tandem mass spectrometry (LC-MS/MS), 4D lipidomics adds an orthogonal, gas-phase physical dimension: ion mobility (1/K0) and calibrated Collision Cross Section (CCS, Ų). This multidimensional architecture directly resolves the single greatest source of false discoveries in untargeted lipidomics: chimeric MS/MS spectra generated by the simultaneous co-fragmentation of co-eluting isomeric and isobaric lipids.

In mammalian tissues, plasma, and cell membranes, thousands of structurally diverse lipid species exhibit overlapping chromatographic retention times and near-identical precursor masses. When a mass spectrometer's quadrupole selection window captures multiple co-eluting precursors, the resulting composite fragmentation spectrum confounds automated search algorithms, leading to phantom lipid annotations, inflated false discovery rates, and distorted quantitative readouts. Through millisecond-scale ion mobility separation synchronized with Parallel Accumulation-Serial Fragmentation (PASEF), 4D lipidomics physically untangles co-eluting precursors prior to collision, delivering pristine, de-chimerized MS/MS spectra that satisfy the most stringent international standards of structural lipid annotation.

Core Analytical Principles of 4D Lipidomics Profiling

Conventional untargeted lipidomics relies on three analytical coordinates: chromatographic retention time (t_R), accurate precursor mass-to-charge ratio (m/z), and tandem mass spectrometric fragmentation (MS/MS). While powerful, this 3D framework frequently collapses when interrogating complex biological matrices. Lipids belonging to the same class share identical polar headgroups and homologous fatty acyl chains that elute in tightly clustered chromatographic windows.

4D Lipidomics integrates an ultra-fast, post-chromatographic separation based on molecular size, shape, and gas-phase charge distribution. As ions emerge from the electrospray ionization (ESI) source, they enter a trapped ion mobility analyzer where they are propelled by a laminar carrier gas flow against an opposing electric field gradient. The fundamental relationship governing this mobility separation is defined by the Mason-Schamp equation:

Where K is the ion mobility, q is ion charge, N is neutral gas number density, k_B is the Boltzmann constant, T is absolute temperature, μ is the reduced mass of the ion-gas collision pair, and Ω is the Collision Cross Section (CCS).

Because the CCS is a true intrinsic physical property determined by the ion's three-dimensional spatial conformation in an inert drift gas (typically molecular nitrogen, N₂), it remains highly reproducible across analytical sequences, instrument platforms, and laboratory sites (≤0.5% RSD). By combining t_R (hydrophobicity and chain length), m/z (elemental composition), MS/MS (connectivity fragments), and CCS (spatial geometry), 4D lipidomics creates a four-dimensional coordinate hyper-cube that reduces identification ambiguity to near zero.

For translational biomarker discovery and comprehensive matrix screening, our validated untargeted lipidomics profiling service leverages high-resolution 4D platforms to deliver audit-ready lipid molecular profiles.

Figure 1. The physical deconvolution of chimeric MS/MS spectra in 4D lipidomics. Left: Conventional 3D LC-MS/MS quadrupole isolation (0.5–1.0 Da) co-selects two co-eluting isomeric lipids, producing an unresolvable chimeric MS/MS spectrum. Right: In 4D TIMS-PASEF, co-eluting precursors are spatially focused by gas-phase shape and sequentially released across milliseconds, enabling the quadrupole to isolate and fragment each isomer independently.

The Chimeric MS/MS Dilemma: Co-Fragmentation and False Over-Annotation

In untargeted lipidomics studies of human plasma or animal tissue extracts, up to 30% to 50% of acquired data-dependent MS/MS spectra are chimeric. A chimeric spectrum occurs when two or more distinct lipid precursors enter the collision cell simultaneously because their precursor m/z values fall within the transmission window of the mass resolving quadrupole (typically 0.4 to 1.0 Da) and their chromatographic peaks overlap.

Mechanisms of Chimeric Interference

• Regioisomeric Co-Elution: Positional isomers differing only in the esterification site of fatty acyl chains on the glycerol backbone (e.g., PC 16:0/18:1 vs PC 18:1/16:0) frequently display nearly identical reverse-phase retention times. When co-fragmented, the resulting spectrum contains contradictory fragment ratios, making accurate quantitative deconvolution impossible.
• Isobaric Lipid Overlap Across Classes: Phosphatidylglycerol (PG) and bis(monoacylglycero)phosphate (BMP/LBPA) share identical elemental formulas and exact masses. Similarly, ether-linked phosphatidylethanolamines (PE O-) share nominal precursor masses with diacyl species from other classes. In a conventional DDA experiment, their co-fragmentation creates chimeric spectra that contain headgroup diagnostic ions from multiple classes.
• Isotopic Envelope Contamination (M+2 Overlap): A monounsaturated lipid species (e.g., containing one double bond, [M+H]⁺) has an isotopic M+2 peak that overlaps with the monoisotopic M+0 peak of a lipid species with one fewer double bond. If the quadrupole window captures both, the fragment ions of the polyunsaturated species contaminate the spectrum of the monounsaturated species, artificially inflating apparent desaturation profiles.

The Severe Consequence of "Phantom" Lipid Identifications

Automated bioinformatic search engines evaluate chimeric spectra by calculating match scores against in silico or experimental spectral libraries. When presented with a chimeric spectrum containing fatty acyl fragments from two co-fragmented lipids (e.g., acyl chains 16:0, 18:1, 18:0, and 20:4), software algorithms often assign the spectrum to a single, fictitious lipid species (e.g., misidentifying a mixture of two diacyl-PCs as a complex multi-acyl lipid or assigning fragments to the wrong precursor). In published clinical cohorts, this failure mode has led to erroneous claims of novel biomarker panels that represent analytical artifacts rather than true disease pathophysiology.

Trapped Ion Mobility Spectrometry (TIMS) & PASEF Synchronization

Trapped Ion Mobility Spectrometry (TIMS) overcomes chimeric co-fragmentation through an ingenious physical mechanism that decouples ion mobility separation from the time of flight. In a TIMS tunnel, an axial gas flow pushes ions toward the mass analyzer, while an opposing electrical field gradient (EFG) holds them stationary at positions where the drag force exactly balances the electrostatic force:

Ions with larger collision cross sections (Ω) experience greater gas drag and are trapped at lower electric field strengths near the front of the tunnel; compact ions with smaller CCS values penetrate deeper into the tunnel before reaching equilibrium. By steadily ramping down the electric field, trapped ions are sequentially eluted in order of decreasing mobility (from large, extended conformations to compact, folded structures) over a cycle time of 50 to 100 milliseconds.

Parallel Accumulation-Serial Fragmentation (PASEF)

The technological breakthrough of 4D lipidomics is the synchronization of the quadrupole mass filter with this millisecond TIMS elution ramp, a process termed PASEF. In standard DDA, the quadrupole can select only one precursor ion per chromatographic elution cycle. In PASEF:

1. During mobility ramp N, ions are trapped and separated in the second TIMS section, while incoming ions for ramp N+1 are accumulated in parallel in the front TIMS section, achieving a nearly 100% ion duty cycle.
2. As ions elute from the TIMS analyzer across the 100 ms window, the quadrupole mass filter rapidly switches its m/z isolation window (switching speeds <1 ms) to match the precursor ions eluting at each specific millisecond of ion mobility.
3. If two isobaric lipid isomers co-elute at t_R = 12.45 min with m/z = 760.58, but possess different gas-phase conformations, Isomer A elutes at 1/K₀ = 1.38 V·s/cm² and Isomer B elutes at 1/K₀ = 1.42 V·s/cm².
4. The quadrupole isolates Isomer A during the first time slice, transmitting it exclusively into the collision cell to acquire a pure, unchimeric MS/MS spectrum. Milliseconds later, the quadrupole isolates Isomer B, generating an independent, uncontaminated spectrum for the second isomer.

Figure 2. Four-dimensional coordinate alignment and tolerance gating. Candidate lipids must satisfy four orthogonal criteria: precursor mass error (Δm/z < 3 ppm), retention time homologous series alignment (ΔtR < 0.2 min), adduct envelope verification, and collision cross section matching (ΔCCS < 1.0%). Any feature violating the multidimensional bounding box is rejected as an analytical artifact.

Multidimensional Coordinate Matching: m/z, Retention Time, Adducts, and CCS

In 4D lipidomics, structural identification is not a simple database search; it is an integrated multi-parameter filtering operation across four independent coordinates:

1. High-Resolution Accurate Mass (m/z)

Precursor ions are measured on a high-resolution time-of-flight (Q-TOF) analyzer with mass accuracy <3 ppm and resolving power >40,000 FWHM. In addition to monoisotopic accurate mass, the experimental isotopic envelope (M+0, M+1, M+2) is compared against theoretical isotopic patterns using metric scoring algorithms (e.g., mSigma score < 20).

2. Chromatographic Retention Time (t_R) Predictability

On reverse-phase C18 or C30 columns, lipids elute according to strict equivalent carbon number (ECN) thermodynamics:

Within any homologous lipid class (e.g., diacyl-PCs), plotting retention time versus carbon chain length yields linear, parallel trajectory lines for species with 0, 1, 2, or 3 double bonds. In 4D lipidomics, any tentative annotation whose experimental t_R deviates from this class-specific retention model is flagged as an isobaric false positive, even if its MS/MS score is high.

3. Cross-Polarity Adduct Envelope Verification

Lipids form multiple adduct species depending on mobile phase ionization chemistry:

• Positive Mode ESI: Phospholipids and glycerolipids form protonated ([M+H]⁺), sodiated ([M+Na]⁺), and ammoniated ([M+NH₄]⁺) ions.
• Negative Mode ESI: Phospholipids form deprotonated ([M-H]^-), acetate ([M+CH₃COO]^-), or formate ([M+HCOO]^-) adducts.

In 4D lipidomics, true biological lipid features must display concordant adduct pairs that share the identical chromatographic retention time (t_R) and demonstrate logical mobility shifts corresponding to adduct mass and size differences. Crucially, in-source decay fragments (such as neutral water losses simulating diacylglycerols from triacylglycerols) lack corresponding intact adduct envelopes, allowing automated algorithms to filter out source-induced artifacts.

4. Collision Cross Section (CCS) Verification

The experimental CCS value serves as the ultimate orthogonal filter. Even if two compounds share identical m/z and similar t_R, an authentic lipid must match its reference CCS within a strict tolerance window (typically ΔCCS < 1.0% to 1.5%).

Collision Cross Section (CCS) Tolerances: Experimental vs. In Silico Libraries

The analytical power of the CCS dimension relies on the quality and calibration of reference collision cross section libraries. In modern 4D lipidomics, researchers utilize two complementary tiers of reference values:

Experimental CCS Calibration Standards

To ensure instrument-independent comparability, TIMS instruments are calibrated prior to analytical queues using certified chemical standards across the entire mass and mobility range (e.g., the Agilent ESI Tune Mix or the Bruker ESI-L Low Concentration Tuning Mix, featuring perfluoroalkylphosphazines with known high-precision drift tube CCS values in N₂).

High-confidence experimental repositories (such as the LipidIMMS database, the AllCCS library, and the Unified CCS Compendium) compile empirically measured CCS values for over 3,000 pure synthetic and natural lipid standards. When matching an experimental feature to an empirical standard measured under identical gas-phase conditions, the maximum acceptable deviation is:

Machine Learning In Silico Predictions (CCSPredict & DeepCCS)

Because the lipidome encompasses hundreds of thousands of theoretical molecular species for which pure synthetic standards do not exist, commercial and open-source platforms deploy advanced deep learning algorithms (such as CCSPredict within MetaboScape or DeepCCS) trained on molecular dynamics and physicochemical descriptors (e.g., number of rotatable bonds, polar surface area, van der Waals volume, dipole moment).

Machine-learning-predicted CCS values typically display a slightly wider median absolute error (1.5% to 2.5%). Consequently, an in silico CCS match is accepted as supportive evidence to prioritize plausible isomeric candidates, but cannot serve as the sole confirmatory criterion for novel biomarker validation.

Figure 3. High-resolution isomeric deconvolution using trapped ion mobility. (A) Extracted ion mobilogram demonstrating baseline separation of the positional regioisomers PC 16:0/18:1 (CCS = 282.4 Ų) and PC 18:1/16:0 (CCS = 286.1 Ų). (B) Pure, unchimeric negative-mode MS/MS spectra acquired via PASEF for each individual mobility peak, definitively resolving sn-1 versus sn-2 carboxylate anion ratios without interference.

Isomeric Deconvolution: Resolving sn-Positional & Geometric Isomers

Lipid isomers represent one of the most challenging frontiers in structural biology. In mammalian metabolism, isomers with identical elemental formulas perform vastly distinct cellular functions:

1. Positional Regioisomers (sn-1 vs. sn-2)

In a diacyl glycerophospholipid, the stereospecific numbering (sn) dictates whether a fatty acid is esterified at the primary hydroxyl (sn-1) or the secondary hydroxyl (sn-2) of the glycerol backbone. Mammalian phospholipases (e.g., PLA₁ vs PLA₂) exclusively cleave one specific bond, releasing signaling messengers like arachidonic acid (20:4) from the sn-2 position.

Because fatty acids at the sn-2 position form a more bent, kinked spatial angle relative to the glycerol backbone than those at sn-1, the two regioisomers possess measurable differences in gas-phase collision cross section (ΔCCS ~ 3 to 6 Ų, or ~1.5% to 2.5%). TIMS instruments operating at resolving powers >60–100 resolve these isomers into distinct mobilogram peaks, allowing independent, de-chimerized fragmentation.

2. Double-Bond Geometric Isomers (cis vs. trans)

Unsaturated fatty acids naturally occur in the cis (Z) conformation, introducing a 30-degree bend into the hydrocarbon chain. Trans-fatty acids (E) adopt an extended, rod-like conformation that packs more tightly, mimicking saturated fats.

In 4D ion mobility, the extended trans isomer presents a distinctly smaller collision cross section than the curved cis isomer. For example, in oleic acid (18:1 9Z) vs elaidic acid (18:1 9E) containing phospholipids, 4D lipidomics resolves the geometric isomers in the gas phase, providing clinical diagnostic platforms with the ability to monitor dietary trans-fat incorporation into cellular membranes.

For programs focusing on phospholipid species distribution, our phospholipids analysis service provides certified structural profiling across complex biological cohorts.

Annotation Confidence Levels: LSI Nomenclature from Sum Formula to Molecular Species

To eliminate chaos in published lipidomic literature, the Lipidomics Standards Initiative (LSI) and the International Lipid Classification and Nomenclature Committee (ILCNC) established strict reporting guidelines (Liebisch et al., 2020). A core mandate in 4D lipidomics is that researchers must never report a level of structural detail that exceeds the actual physical evidence provided by the analytical measurement.

Avoiding the "Over-Annotation" Trap

In many legacy lipidomics publications, authors report lists containing detailed structural strings such as PC 16:0/18:1(9Z) based solely on an automated search engine match against an untargeted Q-TOF run. Under LSI guidelines, this constitutes over-annotation and is scientifically unacceptable:

• If the mass spectrometer detects m/z = 760.58 and confirms the phosphocholine headgroup (m/z 184.07) in positive mode, the correct and only permissible annotation is PC 34:1 (Level 2: Sum Composition).
• If negative-mode tandem MS confirms the presence of both palmitate (16:0) and oleate (18:1) carboxylate anions, the annotation advances to PC 16:0_18:1 (Level 3: Molecular Species), using an underscore _ to indicate that the sn-position is unproven.
• Only when high-resolution ion mobility resolves the isomer into a distinct mobility peak and MS/MS confirms the diagnostic carboxylate intensity ratio (or matches a pure synthetic regioisomeric standard) may the slash / be used: PC 16:0/18:1 (Level 4: Positional Isomer).

Bioinformatic Deconvolution Pipelines: MS-DIAL 5 & MetaboScape Implementations

Processing four-dimensional lipidomics datasets requires advanced bioinformatic algorithms capable of handling massive raw data volumes (typically 1 to 5 gigabytes per LC-MS run) while executing multi-frame 4D feature extraction:

1. 4D Feature Detection & Alignment (MS-DIAL 5)

The open-source MS-DIAL 5 platform natively supports 4D timsTOF and DTIMS data structures. The feature extraction engine identifies ion clusters across all four dimensions simultaneously:

1. Centroiding & Mobilogram Slicing: Extracts ion currents across retention time (t_R), accurate mass (m/z), and drift time (1/K₀), generating a 4D peak spot defined by its bounding coordinates.
2. Mobility-Assisted Deconvolution: Uses Gaussian mathematical modeling to separate co-eluting peaks along the mobility dimension. If two compounds co-elute in the chromatographic dimension (t_R), MS-DIAL deconvolves their shared MS/MS spectrum by weighting fragment ion intensities against the elution profile of each individual mobility peak.
3. Retention Time & CCS Alignment: Uses internal standards and non-linear LOESS regression to align retention times and CCS values across hundreds of analytical batch files.

2. Commercial Workflow Optimization (Bruker MetaboScape)

Bruker's MetaboScape software utilizes the proprietary T-ReX 4D (Time-aligned Region Complete Extraction) algorithm. T-ReX 4D automatically clusters all isotopic peaks, adducts, and in-source fragments belonging to a single chemical entity into a unified "Bucket". By cross-referencing both experimental CCS values and MS/MS fragmentation against the LipidBlast and HMDB 4D spectral libraries, MetaboScape assigns Annotation Quality (AQ) badges (Color-coded from Platinum to Bronze) reflecting LSI confidence tiers.

Technical Troubleshooting: Matrix Suppression, Peak Splitting & False Discovery

Implementing 4D lipidomics in routine cohort screening requires continuous vigilance against technical artifacts:

1. Mobility Peak Splitting Caused by Conformational Protomers

In electrospray ionization, certain lipid species—particularly phosphatidylethanolamines (PE) and sphingomyelins (SM)—can be protonated at multiple distinct basic sites on the headgroup or backbone (e.g., protonation on the phosphate oxygen vs the amino nitrogen). These distinct chemical states are termed protomers. Protomers frequently adopt different gas-phase conformations, producing two distinct ion mobility peaks for a single pure chemical compound. Failing to recognize protomer formation leads automated software to falsely annotate two different lipid isomers. Protomer splitting can be verified by analyzing pure synthetic standards or by modifying mobile phase pH/buffer additives (e.g., switching from formic acid to ammonium acetate).

2. Environmental Drift in Temperature and Gas Density

Because ion mobility is inversely proportional to neutral gas density (N) and temperature (T), ambient room temperature fluctuations can induce drift in measured 1/K₀ values (>1.5%). To maintain ΔCCS tolerances ≤0.5%:

• Ensure laboratory temperature is regulated within ±1.0°C.
• Inject calibrant solution (e.g., ESI-L Tune Mix) at the beginning, middle, and end of every batch run.
• Apply dynamic internal recalibration using known endogenous ubiquitous lipids (e.g., PC 16:0/18:1, experimental CCS = 282.4 Ų) across the sample queue.

3. High-Concentration Mobility Saturation

Injecting excessively concentrated lipid extracts (>1.0 mg/mL) overloads the TIMS analyzer. Space-charge effects inside the trapped ion accumulation region repel incoming ions, causing severe mobility peak broadening, loss of resolving power, and artificial shifts in observed CCS values (>2.0%). Maintain total lipid injection concentrations between 0.05 and 0.2 mg/mL, ensuring that major phospholipid peaks remain within the linear detector response range.

Instrumentation Physics: DTIMS, TWIMS, and TIMS Architecture Comparison

Ion mobility spectrometry encompasses several distinct gas-phase separation technologies. In translational lipidomics, understanding the physics governing each platform dictates how collision cross sections are calibrated and compared across multi-center trials:

1. Drift Tube Ion Mobility Spectrometry (DTIMS)

In DTIMS, ions travel through a stationary drift tube filled with neutral buffer gas under the influence of a uniform, weak electrostatic field. Because the field is linear and static, ion drift velocity ($v_{\text{d}}$) is directly proportional to the electric field strength ($E$). DTIMS is considered the primary reference standard for collision cross section determinations because CCS values can be calculated directly from first principles using the Mason-Schamp equation without requiring external chemical calibration. However, classical DTIMS operates at moderate resolving powers ($R \approx 40$ to $60$) and suffers from lower ion transmission efficiency due to radial diffusion during transit.

2. Traveling Wave Ion Mobility Spectrometry (TWIMS)

TWIMS utilizes electrodynamic ion guides where repeating voltage pulses ("waves") propel ions through a pressurized gas chamber. Smaller ions roll over the waves more frequently, while bulkier ions are pushed forward. Because the electric field is dynamic and non-uniform, TWIMS cannot measure absolute CCS from first principles; it requires multi-point calibration against certified DTIMS reference standards. Recent developments, such as cyclic ion mobility (cIMS), pass ions through a closed loop multiple times, achieving resolving powers exceeding 200–500 FWHM, though acquisition duty cycles are substantially lower than continuous beam analyzers.

3. Trapped Ion Mobility Spectrometry (TIMS)

TIMS inverts the classical DTIMS geometry: instead of ions moving through stationary gas, ions are held stationary against a moving gas stream by an opposing electric field gradient. This design yields three decisive advantages for complex lipidomics:

• Compact Dimension & High Resolving Power: The effective drift path is achieved within a compact 10-cm tunnel, easily generating resolving powers $R > 80$ to $120$ without requiring ultra-long flight paths.
• 100% Ion Utilization Efficiency: Ions are accumulated in a front funnel section while earlier ions are analyzed in the rear tunnel, eliminating the severe transmission losses (>90% discarded ions) inherent to gated drift tubes.
• Synchronized PASEF Operation: The elution rate can be precisely tuned to match quadrupole mass switching, enabling serial fragmentation of multiple precursors per chromatographic peak.

Quantitative Deconvolution Mathematics: The Spectral Contamination Index

To prevent false-positive annotations in clinical cohort studies, bioinformatic algorithms must quantify the exact degree of spectral contamination before attempting library matching. In 4D lipidomics, every precursor isolation window is assigned a Spectral Contamination Index (CI):

Where $I_{\text{target}}$ is the integrated intensity of the intended precursor ion within the quadrupole transmission window, and $I_{\text{interfering}}$ represents the sum of all co-isolated isobaric, isomeric, or isotopic background ions detected across the mobility scan.

In conventional 3D LC-MS/MS, co-eluting lipid isomers regularly yield Contamination Indices between 40% and 80%, rendering the resulting MS/MS spectra unsearchable without massive deconvolution errors. In contrast, by slicing the precursor beam along the millisecond ion mobility dimension, TIMS-PASEF drives the Contamination Index down to <5% for resolved mobility peaks. Bioinformatic software automatically flags and purges any MS/MS spectrum where $\text{CI} > 20\%$, guaranteeing that database match scores reflect authentic single-molecule fragmentation.

Sample Submission & Experimental Planning Guidelines

• Input Biomass Recommendations:
  • Human / Rodent Plasma or Serum: 20 to 50 μL.
  • Cultured Mammalian Cells: 1 × 10⁶ to 5 × 10⁶ cells (washed thoroughly with PBS to remove serum).
  • Homogenized Tissue Biopsies: 10 to 30 mg wet weight.
• Low-Binding Extraction Protocol: Extract lipids using two-phase methyl tert-butyl ether (MTBE) or methanol/isopropanol precipitation. Avoid plastic tubes with slip agents (oleamide/erucamide) that generate intense contaminant peaks in both positive and negative modes.
• Spike Multi-Class Internal Standards: Incorporate stable isotope-labeled standards representing every major lipid class (e.g., the Avanti SPLASH Lipidomix mass spec standard) prior to extraction to enable class-specific recovery and quantitative normalization.
• Storage and Transportation: Lyophilize lipid extracts or store in chloroform/methanol (2:1 v/v) containing 0.01% BHT at -80°C under an inert nitrogen or argon blanket. Ship on dry ice with temperature monitoring.

References:

1. Vasilopoulou, C. G., et al. (2020). Trapped ion mobility mass spectrometry will drive comprehensive lipidomics in clinical research. Nature Communications, 11(1), 331. doi:10.1038/s41467-019-14044-x.
2. Liebisch, G., et al. (2020). Update on LIPID MAPS classification, nomenclature, and shorthand notation for MS-derived lipid structures. Journal of Lipid Research, 61(12), 1539–1555. doi:10.1194/jlr.S120001025.
3. Züllig, T., et al. (2020). High-throughput lipidomics using trapped ion mobility spectrometry and parallel accumulation-serial fragmentation. Analytical Chemistry, 92(21), 14639–14647. doi:10.1021/acs.analchem.0c03160.
4. Tsugawa, H., et al. (2020). A lipidome atlas in MS-DIAL 4. Nature Biotechnology, 38(10), 1159–1163. doi:10.1038/s41587-020-0531-2.
5. Koelmel, J. P., et al. (2020). Lipidomics Standards Initiative: A framework for consensus and transparency in lipidomics. Nature Metabolism, 2(11), 1145–1147. doi:10.1038/s42255-020-00295-2.
6. Hines, K. M., & Xu, L. (2019). Lipid ion mobility spectrometry: Principles, instrumentation, and applications. Mass Spectrometry Reviews, 38(4-5), 346–371. doi:10.1002/mas.21585.
7. Paglia, G., et al. (2015). Ion mobility-derived collision cross section as an additional measure for lipid identification and profiling in clinical lipidomics. Analytical Chemistry, 87(2), 1137–1144. doi:10.1021/ac503893k.