Computer-Assisted Structure Elucidation (CloudMet 2017)
2 likes587 views
The document discusses the challenges and methodologies related to computer-assisted structure elucidation in chemistry, particularly focusing on the analysis of metabolites and constitutional isomers. It highlights the vast number of potential isomers and the limitations of current deterministic methods due to combinatorial explosions. Additionally, it mentions various techniques and algorithms for structural analysis, emphasizing the importance of spectroscopic input in molecular characterization.
![LSD Input Syntax:Advanced
Liu et al., C34H52O6, HRESI-MS (m/z 557.3833 [m+H])
ELIM 3 4
MULT 1 C 2 0
MULT 2 C 3 0
[… 35 more omitted …]
; known carbonyls
BOND 1 36
BOND 7 37
BOND 10 39
CARB L1 ; define list L1 containing all carbons
HETE L2
LIST L3 2 3 6 8 17 18 27 28
LIST L4 9 38 40
PROP L1 1 L2 - ; Every carbon atom
; can carry one or less
; hetero-atoms, but not
; two
PROP L4 0 L3 ; Every oxygen which is
; not an sp2 O has a
; parter from L3 (based
; on a conservative
; chemical shift
; inspection)
COSY 4 5
[… 4 more omitted …]
HMQC 4 4
[… more omitted …]
HMBC 2 4
[… 98 more omitted …]
http://eos.univ-reims.fr/LSD/index_ENG.html](https://image.slidesharecdn.com/casecomputerassistedstructureelucidationcloudmet2017sardinia-170918135218/85/Computer-Assisted-Structure-Elucidation-CloudMet-2017-75-320.jpg)
![Liu et al., C34H52O6, HRESI-MS (m/z 557.3833 [m+H])
InChIKey=ANWFPAAUCGPEBV-MOHJPFBDNA-N InChIKey=YAJAXOAXTCGOQA-HKOYGPOVNA-N](https://image.slidesharecdn.com/casecomputerassistedstructureelucidationcloudmet2017sardinia-170918135218/85/Computer-Assisted-Structure-Elucidation-CloudMet-2017-76-320.jpg)
![Liu et al., C34H52O6, HRESI-MS (m/z 557.3833 [m+H])
InChIKey=ANWFPAAUCGPEBV-MOHJPFBDNA-N InChIKey=YAJAXOAXTCGOQA-HKOYGPOVNA-N
InChIKey of published compound #14 = YAJAXOAXTCGOQA-HKOYGPOVNA-N](https://image.slidesharecdn.com/casecomputerassistedstructureelucidationcloudmet2017sardinia-170918135218/85/Computer-Assisted-Structure-Elucidation-CloudMet-2017-77-320.jpg)
Computer-Assisted Structure Elucidation (CloudMet 2017)
- 4. 1 10 100 1000 10000 100000 1000000 10000000 100000000 Metabolites in Human Metabolites in Microbes Compounds in ChEBI V154 Metabolites in HMDB V3.6 Metabolites in Plants Compounds in ChEMBL V23 Compounds in PubChem V8-2017 Species Metabolomes and How Little We Know 80,000 200,000
- 5. 1 10 100 1000 10000 100000 1000000 10000000 100000000 Metabolites in Human Metabolites in Microbes Compounds in ChEBI V154 Metabolites in HMDB V3.6 Metabolites in Plants Compounds in ChEMBL V23 Compounds in PubChem V8-2017 Species Metabolomes and How Little We Know 80,000 200,000 2,000,000
- 6. In a typical metabolome measurement, less than 40% of the features can be assigned to known compounds. Oliver Fiehn, UC Davis, USA Internal communication
- 10. There are known knowns; there are things we know we know. We also know there are known unknowns; that is to say, we know there are some things we do not know. But there are also unknown unknowns – the ones we don’t know we don’t know. —United States Secretary of Defense, Donald Rumsfeld
- 11. Levels of Confidence Reproduced from: Viant MR, Kurland IJ, Jones MR, Dunn WB (2017) How close are we to complete annotation of metabolomes? Current Opinion in Chemical Biology 36:64–69. doi: 10.1016/j.cbpa.2017.01.001
- 12. How many constitutional isomers are we looking at? C6H6
- 13. The 217 constitutional isomers of C6H6
- 14. How many constitutional isomers are we looking at? C10H16 C13H16O3 C30H48O2
- 15. How many constitutional isomers are we looking at? C10H16 C13H16O3 C30H48O2 24938 constitutional isomers
- 16. How many constitutional isomers are we looking at? C10H16 C13H16O3 C30H48O2 24938 constitutional isomers > 2,000,000,000 constitutional isomers
- 17. How many constitutional isomers are we looking at? C10H16 C13H16O3 C30H48O2 24938 constitutional isomers > 2,000,000,000 constitutional isomers >> 1012 constitutional isomers
- 18. Christoph Steinbeck 10 Constitutional Isomers of C10H16
- 19. Generating molecular spaces: Algorithms for non-redundant generation of molecular graphs
- 20. Generating molecular spaces: Algorithms for non-redundant generation of molecular graphs (Chemical) Space is big. You just won't believe how vastly, hugely, mind-bogglingly big it is. I mean, you may think it's a long way down the road to the chemist's, but that's just peanuts to space. Douglas Adams, The Hitchhiker's Guide to the Galaxy English humorist & science fiction novelist (1952 - 2001)
- 22. Levels of Confidence Reproduced from: Viant MR, Kurland IJ, Jones MR, Dunn WB (2017) How close are we to complete annotation of metabolomes? Current Opinion in Chemical Biology 36:64–69. doi: 10.1016/j.cbpa.2017.01.001
- 23. Computer-Assisted Structure Elucidation with 2D NMR • The only viable way for slightly more complex problems upwards.
- 24. 1D Proton NMR
- 25. 1D Proton NMR
- 26. 1D Proton NMR
- 27. Exploded Pharmacy: HR-MS yields information about elemental composition, such as C10H16 H C C C C C CCC C C H H H H H H H H H H H H H H H
- 28. Experiments: J-Couplings: DEPT • 13C-detekted 1D-Exp. •Number of protons attached to each carbon is coded as signal phase •Combining information from DEPT-135, DEPT-90 and bb- decoupled carbon nmr yields a complete list of carbon fragments in the molecule. DEPT-90 Zoomed region of DEPT-135 DEPT-135 Time required: 1-5 min Acronyms: DEPT: Distortionless Enhancement by Polarization Transfer APT: Attached Proton Test
- 31. After evaluation of DEPT experiment (or multiplicity edited HSQC) heavy atoms are labeled with chemical shifts and number of attached hydrogen atoms. 31.5 ppm 31.3 ppm 38.0 ppm 144.5 ppm 20.9 ppm 26.4 ppm 23.0 ppm 116.1 ppm 47.2 ppm 40.9 ppm
- 32. Experiments: J-Couplings: HSQC alias HMQC, CH-COSY, HETCOR Acronyms: HSQC: Heteronuclear Single Quantum Coherence HMQC: Heteronuclear Multiple Quantum Coherence 1JCH •Cross-Signals in HSQC-diagrams caused by CH-Couplings via one CH-bond(1JCH, ~140 Hz ). •Experiment yields list of pairs of directly bonded carbon and hydrogen atoms.
- 33. 1JCH Experiments: J-Couplings: HSQC alias HMQC, CH-COSY, HETCOR
- 34. Experiments: J-Couplings: HH-COSY Acronyms: DQF-COSY: Double Quantum Filtered COrrelation Spectroscopy •3JHH-Couplings (and a few others, unfortunately) •Proton-rich skeletons might be elucidated just by HHCOSY und HSQC •Problem: Quarternary carbons, heteroatoms in the skeleton 3JHH
- 35. 3JHH gs-COSY without DQ-Filter 2 scans/increment, 128 increments 5 min experiment time Experiments: J-Couplings: HH-COSY
- 36. 3JHH gs-COSY without DQ-Filter 2 scans/increment, 128 increments 5 min experiment time Experiments: J-Couplings: HH-COSY
- 37. Experiments: J-Couplings:TOCSY Acronyms: TOCSY: Total Correlation SpectroscopY HOHAHA (HOmonuclear HArtmann-HAhn) transfer •Correlates all protons in a scalarly- coupled spin system via spin-lock Puls.
- 38. Experiments: J-Couplings: HMBC Acronyms: HMBC: Heteronuclear Multiple Bond Coherence COLOC: COrrelation via LOng range Kopplungen •Cross-signals through scalar couplings between carbon and hydrogen via 2 or 3 bonds (2 JCH/3 JCH , ~8 Hz ). •Problem: 2 JCH/3 JCH -couplings cannot be distinguished. •Problem: 4 JCH/5 JCH -couplings, which cannot be easily distinguished from 2 JCH/ 3 JCH -couplings
- 40. Experiments: J-Couplings-> INADEQUATE Acronyms: INADEQUATE: Incredible Natural Abundance DoublE QUAnTum CoherencE •Powerful 13 C-detected method •Cross signals via 1-bond CC couplings(1 JCC , ~40 Hz ) •Problem: 0.011x 0.011 ~ 1/10000 (0.01%). •Large amount of substance needed (10mg/C-atom)
- 42. Experiments: Dipolar Couplings: NOESY Acronyms: NOESY: Nuclear Overhauser Enhanced SpectroscopY •Cross signals via dipolar couplings through space between protons that are less than 5 Å apart in space •I ~ r-6 •Important experiment for 3D structure determination of macromolecular structure.
- 43. After evaluation of HMBC, we are looking at a molecular puzzle of pairs of carbon atoms that are either 1 or 2 bonds apart (but we don’t know which is the case). 31.5 ppm 31.3 ppm 38.0 ppm 144.5 ppm 20.9 ppm 26.4 ppm 23.0 ppm 116.1 ppm 47.2 ppm 40.9 ppm 1.16 ppm 2.34 ppm 1.63 ppm 2.19 ppm 0.85 ppm 1.27 ppm5.17 ppm 2.06 ppm 1.93 ppm
- 44. Structure Elucidation (CASE): Step by Step HR-MS, et al. Typical 1H and 13C-chemical shifts DEPT, 1JCH-Correlations HH COSY, TOCSY HMBC 1,1-ADEQUATE Gross Formula Functional Groups Structural Fragments Constitution Relative Configuration Complete Structure 3JHH-Couplings NOE-Diff. spectra HH NOESY, HH ROESY C10H16 2x 3x 3x 2x
- 46. Computer-Assisted Structure Elucidation (CASE): Step by Step HR-MS, et al. Typical 1H and 13C-chemical shifts DEPT, 1JCH-Correlations HH COSY, TOCSY HMBC 1,1-ADEQUATE Gross Formula Functional Groups Structural Fragments Constitution Relative Configuration Complete Structure 3JHH-Couplings NOE-Diff. spectra HH NOESY, HH ROESY C10H16 2x 3x 3x 2x
- 47. Yes, it is the same workflow and you need a bit of cheminformatics behind the scene to do the job
- 48. The Chemistry Development Kit (CDK): An Open Source Java Library for Structural Cheminformatics http://cdk.github.io
- 52. Fast structure searches via binning
- 53. 40 years of CASE research
- 56. • Sesami • Assemble • Houdini
- 57. Munk, M.E. et al., 1982. Computer-assisted structure elucidation. Fresenius' Zeitschrift für analytische Chemie, 313(6), pp.473–479.
- 59. LSD (Logic for Structure Determination) • LSD (Logic for Structure Determination • Command-line driven • Takes spectral constraints as input • Generates lists of connection tables (molecules) • Open Source • Rocket-fast • No early spectrum processing
- 60. Christoph Steinbeck 48 Ab-Initio Structure Elucidation by 2D NMR No. 13C CPD 1JCH (HMQC) HMBC 1 148.24 - 4.0; 2.42; 2.28; 1.20 2 118.25 5.49 4.0; 2.28; 2.17 3 66.32 4.0 5.49 4 43.87 2.17 5.49; 4.0; 2.42; 1.32; 1.20; 0.86 5 41.40 2.14 5.49; 1.32; 0.86 6 38.32 - 2.42; 1.32; 1.20; 0.86 7 32.00 1.20/2.42 8 31.52 2.28 2.42; 1.20 9 26.52 1.32 0.86; 2.17 10 21.46 0.86 1.20; 1.32 2D peak picking table Table with heavy-atom relations Internally generated by CASE program Steinbeck, C. Computer-Assisted Structure Elucidation. In Handbook on Chemoinformatics.; Gasteiger, J. Ed.; Wiley-VCH: Weinheim, 2003; Vol. 2; pp. 1378-1406.
- 61. Limits to Growth
- 62. Limits to Growth • Deterministic methods suffer from combinatorial explosion
- 63. Limits to Growth • Deterministic methods suffer from combinatorial explosion No. of Heavy Atoms No.ofConstitutionalIsomer CalculationTime
- 64. Limits to Growth • Deterministic methods suffer from combinatorial explosion No. of Heavy Atoms No.ofConstitutionalIsomer CalculationTime C10H16 (10 Heavy Atoms) 24938 Constitutional Isomers
- 65. Limits to Growth • Deterministic methods suffer from combinatorial explosion No. of Heavy Atoms No.ofConstitutionalIsomer CalculationTime C13H16O3 (16 Heavy Atoms) > 2,000,000,000 Constitutional Isomers O O O C10H16 (10 Heavy Atoms) 24938 Constitutional Isomers
- 66. Limits to Growth • Deterministic methods suffer from combinatorial explosion No. of Heavy Atoms No.ofConstitutionalIsomer CalculationTime C13H16O3 (16 Heavy Atoms) > 2,000,000,000 Constitutional Isomers O O O C10H16 (10 Heavy Atoms) 24938 Constitutional Isomers C30H48O2 (32 Heavy Atoms) >> 1012 Constitutional Isomers HO OH
- 67. Limits to Growth • Deterministic methods suffer from combinatorial explosion • Prospective use of spectroscopic input information may make them error-intolerant No. of Heavy Atoms No.ofConstitutionalIsomer CalculationTime C13H16O3 (16 Heavy Atoms) > 2,000,000,000 Constitutional Isomers O O O C10H16 (10 Heavy Atoms) 24938 Constitutional Isomers C30H48O2 (32 Heavy Atoms) >> 1012 Constitutional Isomers HO OH
- 68. Deterministic Structure Generators: The LUCY Method • Prospective use of spectral information for building isomers • Needs 1D 13C, 2D HMQC, HMBC, HH COSY • Example: Walk a decision tree while interpreting HMBC-Signals N Atoms, No Bonds Inserted Bonds: 1 2 3 n HMBC-derived relations between Heteroatoms Steinbeck, C.; Angewandte Chemie. International Ed. in English 1996, 35, 1984-1986.
- 69. LSD Input Syntax: Basics ⍺-Pinene Example http://eos.univ-reims.fr/LSD/index_ENG.html MULT 1 C 2 0 MULT 2 C 2 1 MULT 3 C 3 1 MULT 4 C 3 1 MULT 5 C 3 0 MULT 6 C 3 2 MULT 7 C 3 2 MULT 8 C 3 3 MULT 9 C 3 3 MULT 10 C 3 3 HMQC 2 2 HMQC 3 3 HMQC 4 4 HMQC 6 6 HMQC 7 7 HMQC 8 8 HMQC 9 9 HMQC 10 10 HMBC 1 6 HMBC 1 9 HMBC 2 3 HMBC 2 9 HMBC 3 6 HMBC 3 8 HMBC 3 9 HMBC 3 10 HMBC 4 6 HMBC 4 8 HMBC 4 10 HMBC 5 6 HMBC 5 8 HMBC 5 10 HMBC 7 6 HMBC 8 10 HMBC 9 3 HMBC 10 8 Atomdefinitions1JCH,redundant? 2/3JCH,C-H-long-rangecorrelations C oncatenate those blocks in a textfile
- 70. LSD Usage ⍺-Pinene Example $ lsd pinene.in $ outlsd 7 < pinene.sol > pinene.sdf http://eos.univ-reims.fr/LSD/index_ENG.html
- 71. LSD Usage ⍺-Pinene Example $ lsd pinene.in $ outlsd 7 < pinene.sol > pinene.sdf http://eos.univ-reims.fr/LSD/index_ENG.html Command prompt - don’t type this in
- 72. LSD Usage ⍺-Pinene Example $ lsd pinene.in $ outlsd 7 < pinene.sol > pinene.sdf http://eos.univ-reims.fr/LSD/index_ENG.html Command prompt - don’t type this in will produce pinene.sol (LSD specific output file)
- 73. LSD Usage ⍺-Pinene Example $ lsd pinene.in $ outlsd 7 < pinene.sol > pinene.sdf http://eos.univ-reims.fr/LSD/index_ENG.html Command prompt - don’t type this in will produce pinene.sol (LSD specific output file) converts .sol into .sdf (standard molecular file format) Use e.g. MarvinView to view .sdf
- 74. LSD Usage ⍺-Pinene Example MarvinView rendering of the two results in accordance with out input data from the previous slide
- 75. LSD Input Syntax:Advanced Liu et al., C34H52O6, HRESI-MS (m/z 557.3833 [m+H]) ELIM 3 4 MULT 1 C 2 0 MULT 2 C 3 0 [… 35 more omitted …] ; known carbonyls BOND 1 36 BOND 7 37 BOND 10 39 CARB L1 ; define list L1 containing all carbons HETE L2 LIST L3 2 3 6 8 17 18 27 28 LIST L4 9 38 40 PROP L1 1 L2 - ; Every carbon atom ; can carry one or less ; hetero-atoms, but not ; two PROP L4 0 L3 ; Every oxygen which is ; not an sp2 O has a ; parter from L3 (based ; on a conservative ; chemical shift ; inspection) COSY 4 5 [… 4 more omitted …] HMQC 4 4 [… more omitted …] HMBC 2 4 [… 98 more omitted …] http://eos.univ-reims.fr/LSD/index_ENG.html
- 76. Liu et al., C34H52O6, HRESI-MS (m/z 557.3833 [m+H]) InChIKey=ANWFPAAUCGPEBV-MOHJPFBDNA-N InChIKey=YAJAXOAXTCGOQA-HKOYGPOVNA-N
- 77. Liu et al., C34H52O6, HRESI-MS (m/z 557.3833 [m+H]) InChIKey=ANWFPAAUCGPEBV-MOHJPFBDNA-N InChIKey=YAJAXOAXTCGOQA-HKOYGPOVNA-N InChIKey of published compound #14 = YAJAXOAXTCGOQA-HKOYGPOVNA-N
- 79. Stochastic Search Methods • Simulated Annealing • Traveling Salesman Problem • Finding the solution structure of large biomolecules • Integrated Circuits Layout • Robotic Path Planning • Genetic Algorithms • Protein Folding • Immune System Simulation • Computer-Aided Design • Quite a number of other options ... Algorithms known to tackle large search spaces:
- 80. Simulated Annealing Guided Walk in Constitution Space
- 81. Simulated Annealing Guided Walk in Constitution Space
- 82. Simulated Annealing Guided Walk in Constitution Space
- 83. Simulated Annealing Guided Walk in Constitution Space
- 84. Simulated Annealing Guided Walk in Constitution Space Neighbors in constitution space
- 85. Simulated Annealing Guided Walk in Constitution Space Neighbors in constitution space • are in close chemical distance to each other
- 86. Simulated Annealing Guided Walk in Constitution Space Neighbors in constitution space • are in close chemical distance to each other • are likely to be similar in their spectroscopic properties
- 87. Simulated Annealing Guided Walk in Constitution Space Neighbors in constitution space • are in close chemical distance to each other • are likely to be similar in their spectroscopic properties
- 88. Simulated Annealing Small steps on the constitution space landscape Faulon, J.-L.; J. Chem. Inf. Comput. Sci., 36 (1996) 4, 731-40
- 89. Simulated Annealing Evaluating a score for each point (constitution) in structure space Score Function based on Spectroscopic Fitness Stotal = c1 SHMBC + c2 SHHCOSY + c3 SShift + … + cn SFeatures
- 90. Simulated Annealing Small steps on the constitution space landscape exp( ) f p T δ = − T t0,0 Annealing Schedule Tt=αTt-1 with 0.9 < α < 1 Acceptance criterion
- 91. Simulated Annealing Small steps on the constitution space landscape exp( ) f p T δ = − T t0,0 Annealing Schedule Tt=αTt-1 with 0.9 < α < 1 Acceptance criterion
- 92. Simulated Annealing Small steps on the constitution space landscape exp( ) f p T δ = − T t0,0 Annealing Schedule Tt=αTt-1 with 0.9 < α < 1 Acceptance criterion
- 93. Simulated Annealing Small steps on the constitution space landscape exp( ) f p T δ = − T t0,0 Annealing Schedule Tt=αTt-1 with 0.9 < α < 1 Acceptance criterion
- 94. Some Nice Properties of this SA Scheme
- 95. Some Nice Properties of this SA Scheme •Pluggable Target Function
- 96. Some Nice Properties of this SA Scheme • If you can reliably calculate a measurable property for a given constitution, it can be part of your target function •Pluggable Target Function
- 97. Some Nice Properties of this SA Scheme • Spectroscopic information • IR • UV-VIS • Other types of NMR experiments • MS fragmentation (?) • If you can reliably calculate a measurable property for a given constitution, it can be part of your target function •Pluggable Target Function
- 98. Some Nice Properties of this SA Scheme • Spectroscopic information • IR • UV-VIS • Other types of NMR experiments • MS fragmentation (?) • If you can reliably calculate a measurable property for a given constitution, it can be part of your target function •Pluggable Target Function • Additional knowledge • Good-List/Bad-List fragments • Drug Likeness • Natural Product Likeness
- 99. Some Nice Properties of this SA Scheme • Spectroscopic information • IR • UV-VIS • Other types of NMR experiments • MS fragmentation (?) • If you can reliably calculate a measurable property for a given constitution, it can be part of your target function •Pluggable Target Function • Additional knowledge • Good-List/Bad-List fragments • Drug Likeness • Natural Product Likeness General System for Optimization in Constitution Space
- 100. Some Nice Properties of this SA Scheme • Spectroscopic information • IR • UV-VIS • Other types of NMR experiments • MS fragmentation (?) • If you can reliably calculate a measurable property for a given constitution, it can be part of your target function •Pluggable Target Function •Artifacts only lead to slightly lower ranking of correct structure in hit list • Additional knowledge • Good-List/Bad-List fragments • Drug Likeness • Natural Product Likeness General System for Optimization in Constitution Space
- 101. Score convergence for SENECA SA run on Polycarpol (32 heavy atoms), performed by 8 server processes Score 3000 4250 5500 6750 8000 Iteration 0 33500 67000 101000134500 168000201500 235000268500 302000335500 HO OH Polycarpol (C30H48O2). Distributed computing at its cheapest. Distributed Server Client Gatekeeper Retrieve Server List Collect ResultsSubmit Spectral Data
- 102. Qualitative assessment: Computational complexity of deterministic and stochastic algorithms Compound LUCY SENECA Steps overall α-Pinene (C10H16) 2 s 1 min 30,000 Eurabidiol(C15H28O2) 29 s 5 min 90 000 Polycarpol (C30H50O) 33 min 12 min 350,000 OH HO HO OH Eurabidiol (C15H28O2) Polycarpol (C30H48O2).α-Pinene (C10H16) Deterministic vs SA Generation Timeinseconds 0 500 1000 1500 2000 No of Heavy Atoms 0 8 15 23 30 Deterministic SA C. Steinbeck, Journal of Chemical Information & Computer Sciences 2001, 41, 1500.
- 103. Ranking Solutions • Heteroatom-rich/proton-poor skeletons can yield many solutions (hundreds, thousands) • Possible ranking by • Spectrum Similarity • Natural Product Likeness
- 104. CONTENTS
- 106. Components for Molecule Curation
- 107. Components for Molecule Curation
- 118. Component for Signature Generation
- 121. NP - Natural product SM - Synthetic molecule Statistics for Scoring
- 122. In the fragment contribution (Fragmenti), ✦NPi is the total number of molecules in the NP dataset in which the Fragmenti occurs, ✦SMi is the total number of molecules in the SM dataset in which the Fragmenti occurs, ✦SMt is the total number of molecules int he SM dataset ✦NPt is the total number of molecules in the NP dataset. ✦N is the number of fragments in given molecule NP - Natural product SM - Synthetic molecule Statistics for Scoring
- 123. In the fragment contribution (Fragmenti), ✦NPi is the total number of molecules in the NP dataset in which the Fragmenti occurs, ✦SMi is the total number of molecules in the SM dataset in which the Fragmenti occurs, ✦SMt is the total number of molecules int he SM dataset ✦NPt is the total number of molecules in the NP dataset. ✦N is the number of fragments in given molecule NP - Natural product SM - Synthetic molecule Statistics for Scoring
- 124. In the fragment contribution (Fragmenti), ✦NPi is the total number of molecules in the NP dataset in which the Fragmenti occurs, ✦SMi is the total number of molecules in the SM dataset in which the Fragmenti occurs, ✦SMt is the total number of molecules int he SM dataset ✦NPt is the total number of molecules in the NP dataset. ✦N is the number of fragments in given molecule NP - Natural product SM - Synthetic molecule Statistics for Scoring
- 125. Jayaseelan KV, Moreno P, Truszkowski A, Ertl P & Steinbeck C (2012) Natural product-likeness score revisited: an open-source, open-data implementation. BMC Bioinformatics 13, 106. • Natural Product-likeness classification and integrated it into Taverna workflow tool • (http://sourceforge.net/projects/np-likeness/). • Included in second version of SENECA CASE
- 126. Availability
- 127. Publicationen
- 128. Publicationen
- 129. Publicationen