a novel algorithm for automating fragment ion structure assignment ... · a novel algorithm for...

PO-CON1885E

A novel algorithm for automatingfragment ion structure assignmentusing high mass accuracy MS/MS data

ASMS 2019

Neil Loftus1, Kirsten Hobby1, Alan Barnes1; 1Shimadzu Corporation, Manchester, UK.

2

A novel algorithm for automating fragment ion structure assignment using high mass accuracy MS/MS data

Overview• To help the interpretation of accurate mass MS/MS spectra and to provide an effective means of structurally verifying

component identi�cation a novel algorithm has been developed to automate fragment ion annotation. • The algorithm has been applied to high resolution accurate mass (HRAM) fragment ion data for a panel of chemically

diverse small molecules. This test panel generated a wide variety of fragmentation mechanisms including single bond heterolytic and homolytic cleavages, ring opening and cleavage, and intramolecular rearrangements resulting in neutral losses such as ammonia, carbon monoxide and hydrogen cyanide.

IntroductionHigh mass accuracy mass spectrometry platforms are powerful analytical tools generating accurate mass measurement information and product ion spectra applied to component identi�cation and/or veri�cation. However, a key challenge in process relates to interpreting product ion spectra and assigning chemical structures to speci�c fragment ions. For the computational prediction and annotation of product ion spectra a number of methods

have been proposed including input–output kernel regression machine learning spectra and bond dissociation approaches combined with data base searches. In this paper, a novel algorithm has been developed to assign structures to fragment ions to help fragment interpretation and compound veri�cation but also as a tool to construct accurate mass fragment ion data for quantitative and qualitative Q-TOF analysis.

Materials and MethodsA panel of small molecules was used to test the algorithm in this research application, including pesticides extracted from complex food matrices using a standard QuEChERS protocol, drugs of abuse spiked into plasma and plasma crash samples from a metabolomics study. Data was acquired with the LCMS-9030 QTOF system (Shimadzu Corporation). The acquisition method design was typically set-up with a MS and DIA-MS/MS. HRAM Q-TOF (LCMS-9030 Shimadzu Corporation) data acquired sequential mass scans with a cycle time less than 1 second (MS and subsequent DIA-MS/MS mass scans with an isolation width of m/z 20). In each MS/MS event a collision energy (CE) spread was applied to generate precursor MS and product ion MS/MS spectra.

The algorithm functions by identifying groups of atoms that could be involved in a fragmentation mechanism, then uses the fragment ion data to con�rm each putative mechanism. The �rst step in the process is the

consideration of tautomeric shifts within the ionized precursor molecule. Collections of ionized tautomer molecules are created as the starting point in the process. Each charged tautomer is then analyzed for the presence of groups of atoms that could be involved in a putative mechanism. Next the rearrangement of electrons is considered with various outcomes postulated, such as homolytic or heterolytic cleavage or the creation of new bonds that did not originally exist. Molecular charge and chemical constraints are then evaluated, and only valid hypotheses accepted. Finally, iterative combinations of the hypothetical fragmentation mechanisms are then considered to give fragment ions that are formed by multistage mechanisms. Using acquired LCMS-9030 Q-TOF data, product ion spectra were annotated and used to build accurate mass methods and verify compound identi�cation. Figures 1-3 highlight suggested fragments based on original structures.

3


Targeted and untargeted analysis food safety

Results

The algorithm has been applied to annotate fragment structures to verify component identi�cation and distinguish isomers in complex matrices. In the example below the research algorithm has been applied to DIA-MS/MS spectra of ethiofencarb and methiocarb following HRAM analysis.

Figure 1. Automatic structural assignment for fragment ions of ethiofencarb and methiocarb. The DIA-MS/MS mass spectra differentiate the components based upon fragment ions at speci�c ions at m/z 121 for methiocarb and m/z 164 for ethiofencarb (ethiofencarb’s loss of carbamate group leads to formation of ion at m/z 169 that subsequently leads to ion at m/z 121 after losing methanethiol; it is likely that the hydroxybenzyl cation at m/z 164 could rearrange to hydroxytropyllium ion (Nunez at al, Toxics 2018)).

Assigning fragment structures to ethiofencarb and methiocarb in a complex food extractQuEChERS extract spiked with over 200 target pesticidesUntargeted MS and DIA-MS/MS; QTOF Data Acquisition LCMS-9030Cycle time 0.8 second for all MS and DIA-MS/MS mass scans MS mass scan; mass range 140-900 Da (40 msecs)MS/MS 39 sequential mass scans; mass range 65-900 Da (20 msecs for each mass scan)Precursor isolation width 20 m/z; CE spread 0-30VExternal mass calibration

DIA-MS/MS Spectrum; MethiocarbFragment structures assigned

Ethiofencarb RT 5.401 minsC11H15NO2SCAS 29973-13-5Molecular ion 226.08963 (M+H)+

Methiocarb RT 6.296 minsC11H15NO2SCAS 2032-65-7Molecular ion 226.08963 (M+H)+

RT (min)2 3 4 5 6 7 8 9 10 11

DIA-MS/MS Spectrum; EthiofencarbFragment structures assigned

79.0539 89.0418

107.0492

164.0706 169.0683226.0899

m/z70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250

O

S

NH

O

O

NH

O

OH

O

S

C7H7Om/z 107.04914

0.6 ppm

C9H10NO2m/z 164.07060

0.0 ppm

C9H13OSm/z 169.06816

0.8 ppmEthiofencarbC11H16NO2Sm/z 226.08963

1.2 ppm

77.03805 93.06977

121.06476

122.07154

169.06826

226.08997

m/z

70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250

Methiocarb C11H16NO2Sm/z 226.08963

1.5 ppm

O

NH

O

S

OHSHOH

C9H13OSm/z 169.06816

0.6 ppm

C8H9Om/z 121.06479

0.3 ppm

4


Fragment ion annotation in DoA screeningA panel of drugs of abuse (DoA) was spiked into human plasma and extracted using QuEChERS. As one example, to verify the identi�cation of 2 selected drugs the algorithm was used to annotate the DIA-MS/MS spectra for fentanyl and cocaine in the test panel.

Assigning fragment structures to fentanyl and cocaine in a human plasma extract spiked with a panel of drugs of abuse (DoA) at 5 ng/mL (precursor mass chromatogram ±5ppm)Untargeted MS and DIA-MS/MS; QTOF Data Acquisition LCMS-9030Cycle time 0.4 second for all MS and DIA-MS/MS mass scans MS mass scan; mass range 100-500 Da (20 msecs)MS/MS 19 sequential mass scans; mass range 40-500 Da (20 msecs for each mass scan)Precursor isolation width 20 m/z; CE spread 0-30VExternal mass calibration

Figure 2. Assignment of DIA-MS/MS mass spectra for fentanyl and cocaine in a human plasma extract using a collision energy spread of 0-30V.

Fentanyl RT 6.701 minsC22H28N2OCAS 437-38-7Molecular ion 337.22744 (M+H)+

RT (min)1 2 3 4 5 6 7 8 9 10

Cocaine RT 5.366 minsC17H21NO4CAS 50-36-2Molecular ion 304.15433 (M+H)+

DIA-MS/MS Spectrum; FentanylFragment structures assigned

105.0697

132.0804

188.1433

216.1381

m/z40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340

C13H18Nm/z 188.14338

0.4 ppmC9H10N

m/z 132.08078 2.9 ppm

C8H9m/z 105.06988

1.7 ppm

337.2275N

N

O

CH3

N

O

CH3

N

Fentanyl C22H29N2Om/z 337.22752

0.2 ppm

C14H18NOm/z 216.13829

0.9 ppm

N

N

C9H12Nm/z 134.09643

3.0 ppm

C9H12NOm/z 150.09134

3.6 ppm

105.0332 122.0964 150.0908

182.1175

C10H16NO2m/z 182.11756

0.3 ppm

304.1544

305.1575

306.1598

80 100 120 140 160 180 200 220 240 260 280 300

N

CH3

O

O

H H

N

CH3

O

H H

C7H5Om/z 105.03349

2.8 ppmO

N

CH3

H H

C8H12Nm/z 122.09643

0.2 ppm

N

CH3

O

OH3C

O

O

H H

DIA-MS/MS Spectrum; CocaineFragment structures assignedCocaine

C17H22NO4m/z 304.15433

0.2 ppm

5


Figure 3. DIA-MS/MS mass spectrum annotation for lysophosphatidylcholine 16:0 including the characteristic product ion at m/z 104 to identify the sn-1 isoform.

Component veri�cation in metabolomic studiesOne of the key research areas for verifying component identi�cation is in metabolomics and particularly with lipid species. The fragment ion annotation algorithm was applied to structural veri�cation of lysophosphatidylcholine 16:0.

RT (min)7.2 7.4 7.6 7.8 8.0 8.2 8.4 8.6

C24H51NO7P496.33872.1 ppm

m/z50 100 150 200 250 300 350 400 450 500

m/z50 100 150 200 250 300 350 400 450 500

C24H49NO6P478.32900.4 ppm

C19H37O3313.27244.2 ppm

184.0730

124.9992

184.0729

313.2734478.3285

496.3397

DIA-MS/MS Spectrum; Lysophosphatidylcholine16:0Fragment structures assigned

PO

OO

O

OHO

O

N

PO

OHO

O

N

C5H15NO4Pm/z 184.07332

1.7 ppm

LPC 16:0C24H51NO7Pm/z 496.33977

ON

C5H14NOm/z 104.10699

2.8 ppm

PO

OHO

O

C2H6O4Pm/z 124.99982

2.6 ppm

124.9995

sn-2LPC-16:0

Rt 7.7 mins

sn-1LPC-16:0

Rt 8.1 mins

sn-1DIA-MS/MSSpectrum

sn-2DIA-MS/MSSpectrum

H

H

H

104.1067


First Edition: July, 2019

© Shimadzu Corporation, 2019

Conclusions• The fragment ion data for a panel of chemically diverse small molecules generated a wide variety of fragmentation

mechanisms including single bond heterolytic and homolytic cleavages, ring opening and cleavage, and intramolecular rearrangements resulting in neutral losses such as ammonia, carbon monoxide and hydrogen cyanide. Using acquired LCMS-9030 Q-TOF data, product ion spectra were annotated and used to build accurate mass methods and verify compound identi�cation.

• Automated correlation of experimental fragmentation spectra with hypothetical fragment ion structures.• The research application was used to verify component identi�cation in complex matrices using DIA-MS/MS mass spectra

acquired using cycle times less than 1 second (typically for a mass range of 100-1000 Da, MS and 38 subsequent DIA-MS/MS mass scans were acquired with an isolation width of m/z 20). In each MS/MS event a collision energy spread of 0-30V was applied to generate product ions spectra.

a novel algorithm for automating fragment ion structure assignment ... · a novel algorithm for...

Documents