Physicochemical property based scoring scheme for design of an aminerg GPCR-targeted fragment library: The Fragment-GPCR-Score „FrAGS” UGM Budapest 21st of may 2014 Adam A.
Download ReportTranscript Physicochemical property based scoring scheme for design of an aminerg GPCR-targeted fragment library: The Fragment-GPCR-Score „FrAGS” UGM Budapest 21st of may 2014 Adam A.
Slide 1
Physicochemical property based scoring scheme for
design of an aminerg GPCR-targeted fragment library:
The Fragment-GPCR-Score
„FrAGS”
UGM Budapest 21st of may 2014
Adam A. Kelemen
Hungarian Academy of Sciences
Medicinal Chemistry Research Group
Slide 2
INTRODUCTION:
G-PROTEIN COUPLED RECEPTORS AND THE FRAGMENT-BASED
DRUG DISCOVERY
Fragment-based approach:
Aims of the study:
Biophysical or biochemical screening of small-sized, simple, polar molecules in high-concentrations
Advantages:
1. Better sampling of the chemical space
2. Freedom in optimizing them to lead-molecules with favourable pharmacokinetic properties
Establishment of a scoring scheme supporting the design of GPCR-targeted fragment libraries
Characteristic:
Screening at higher concentrations
Rules for the scoring scheme are derived through:
Fragment-library design:
1. Scaled
to the
fragment-screening
1. data
mining
of an open paradigm
access bioactivity database (ChEMBL)
2. Diversity (fragment chemical space: 107)
using ChemAxon’s Instant Jchem and Knime
G-protein coupled
biology,
2.receptors,
Examination
of localization:
physicochemical descriptors
1. Largest cell-membrane receptor family
(well known from Rule of Three, Rule of Five)
2. Endogenous ligands: neurotransmitters, peptides, hormones
using
ChemAxon’s
for Excel
3. 7TM
structure,
extracellularJChem
N-terminal
4. 85% of GPCRs belongs to class A:
(aminergic-, chemokine-, glycoprotein-hormone-, neuropeptide-)
5. Approx. 40% of marketed small molecule drugs are GPCR-targeted
Stephen P. Andrews, ChemMedChem, 2014, 9, 256-275
Slide 3
DATA COLLECTION AND PREPARATION:
TRAINING SETS
Reference
(inactive)
set
Active
set
GPCRs
Class A
ChEMBL GPCR SARfari
(EBI,
EMBL),(active
Data mining
with Instant
Entire
ChEMBL
set subtracted)
AminergMiner
Jchem and Konstanz Information
947914 entries: molecules,
in vitro-essay data1292344
(binding,entry
functional, ADME), GPCR-target
1. Muscarinic Achetylcholine, ACM[x]
2. Adrenoceptors, α[x], β[x]
Dopamine, DRD[x]
1. Activity data: binding essay, IC50, logIC3.
50, Ki, logKi conversion to pKi
1. Counter ions ofdata
salts stripped 4. Histamine, HRH[x]
2. species-independent
10477 fragments
309962 fragments
5000 random sampling
5. Serotonine, 5HT[x]
2. 8 ions
≤ Nheavy
≤ 22
3. Counter
of salts
stripped
6. Octopamine
4. 8 ≤ Nheavy ≤ 22
7. Trace Amine, TAAR[x]
Rate of aminergic-data
in the active set
derived
from ChEMBL GPCR-SARfari (2370 fragments):
Size-independent
ligand
efficiency:
187 fragments contain𝑝𝐾
at least 4 active data on only not-aminergic class-A receptors
2370 active fragments
𝑆𝐼𝐿𝐸 = [ 𝑖
0,3 ] ≥ 1.95 on at least 4 GPCR-receptors
Remaining 2183 fragments𝑁ℎ𝑒𝑎𝑣𝑦
possess activity data only on aminergic GPCRs
2183 fragments with related activity on aminergic GPCR’s
Slide 4
RED: GPCR fragments
BLUE: inactive reference set
CALCULATION OF PHYSICOCHEMICAL PROPERTIES WITH JCHEM
FOR EXCEL
logP (at pH = 7.4)
Polar Surface Area
0.2
4. Rotatable Bond Count (nROT),
0
0
40
2
3
4
80
100
120
140
logP
Acceptor
Count (HBA)
PSA
60
5.
1
-2
0
0
0
0.6
0.15 0.150
0.1 0.100
Frequency
0.58.
0.2 0.200
Number of Nitrogen Atoms
0.4
9. Number of Oxygen Atoms
0.3
0.050
0.2
0.000
0
1
1
2
2 0.1
3
4
5
3
4
5
Number of Nitrogen Atoms
0 HBA
Base
6
6
Acid
7
7
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
02
31
4 2
5
logD
Rotatable Bond Count
3
6
4 7
5
Number of Oxygen Atoms
Donor Count
0.400
0.350
0.300
Frequency
0.25 0.250
-1 1
BA character
7. Strongest basic pKa
Frequency
Frequency
5
0.8
0.350
0
0.05
0.05
6. Strongest acidic pKa
0.400
0.7
0
160
0.1
0.1
Number of
Nitrogen Atoms
0.9
Acceptor
Count
0.3 0.300
0.05
0.2
0.15
3. Polar Surface Area (PSA)
0.05
20
0.25
Frequency
Frequency
0.1
-1
0.25
2. Octanol:water distribution coefficient
0.2 (logD, pH = 7.4)
0.15
0.15
0
0.3
1. Octanol:water partition coefficient (logP)
0.25
6. Donor Count (HBD)
0.35
logD Bond
(at pHCount
= 7.4)
Rotatable
0.3
Frequency
Frequency
0.5
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
-2
0.250
0.200
0.150
0.100
0.050
0.000
0
Zwitterion
0
1
1
2
Neutral
2
3
4
5
3
4
5
Number
of Oxygen
Atoms
HBD
6
6
7
7
Slide 5
RESULTS I. – SELECTION OF GPCR CHARACTERISTIC DESCRIPTORS
1.
Polar Surface Areas (at pH = 7.4) of the active fragments have their median and mean at lower values,
than the reference set.
2.
logP, Donor Count and Acceptor Count distributions do not show significant difference between the
compared sets, however the lower scores of the logD (calculated at pH = 7.4) showed, that GPCR-like
fragments have mostly basic character.
3.
83% of the active fragments resulted to have basic character, 16% of the fragments were neutral.
Aminergic GPCR-active fragments constitute a set of:
I. small sized
II. rigid molecules,
III. containing few heteroatoms,
IV. that are mostly (~83%) basic nitrogens.
Slide 6
RESULTS II. – DESIRABILITY FUNCTIONS AND SCORING
(FRAGMENT GPCR SCORE)
Rotatable Bond Count - SCORE
The desirability function maps the value of a
10 20 30 40 50 60 70 80 90 100 110 120 130 140
TPSA
0.1
0.05
1
2
3
4
5
6
7 8
pKa
9
10 11 12 13 14
7
8
1
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0.9
0.8
0.7
0
0.6
0.5
0.4
0.3
1
2
3
#O
4
5
6
0.2
Number
of Nitrogen Atoms- SCORE
logD (pH = 7.4) – SCORE
1
0.9
0.8 50 60 70 80 90 100110120130140150
0 10 20 30 40
0.7
0.6
Descriptor
0.5
0.4
0.3
0.2
0.1
0
-3
-2
-1
0
1
2
3
4
5
logD
SCORE
0
FrAGS =
[0; 1].
Desirability Function
Number of Oxygen Atoms- SCORE
SCORE
SCORE
Frequency
0.5
1
0.45
0.9
0.8
0.70.4
0.6
0.35
0.5
0.4
0.30.3
0.2
0.1
0.25
0
0.20
0.15
Distribution of the descriptor
pKa - SCORE
SCORE
0
1
0.9
0.8
0.7
0.6
property
onto a score in the range of
0.5
0.4
0.3
0.2
0.1
0
0
1
2
3
4
5
6
nrot
SCORE
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
SCORE
SCORE
PSA – SCORE
6
0.1
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Descriptor
0
1
2
𝑑𝑒𝑠𝑖𝑟𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛 (𝑝𝑟𝑜𝑝𝑒𝑟𝑡𝑦)
3
#N
4
5
6
Slide 7
VALIDATION – CHEMBL, PUBCHEM, FS, HTS
1. ChEMBL validation set
3. Richter Gedeon experimental validation set
High-throughput and fragment screening data on aminergic GPCR targets
2. PubChem validation sets
class-A targeted HTS confirmed actives with at least 10 μM activity
- screenings on allosteric-bindings
- targeting β-arrestin pathways
- or used as counter-screening were sorted out
Targets represented in HTS campaigns:
5HT1A and 5HT1E with 31 confirmed active fragments
TAAR1 with 169 confirmed active fragments
Slide 8
VALIDATION – CHEMBL, PUBCHEM, FS, HTS
ROC
Richter FS
ROC
ChEMBL
6
2.5
5
2
4
0.5
2
1
0
0
0
0.50
-1
1 1.5
0.5
1 1.52 22.5 2.5 3 3 3.53.5 4 4 4.5
4.5
Fragment
GPCR
Score
Fragment
GPCR
Score
55
5.5 6 6
5.5
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0.2 0.3 0.4 0.5 0.6 0.7 0.8
False Positive Rate
False Positive Rate
0.9
ROC
PubChem
Richter
5HT
HTSROC
1
PubChem
TAAR1 ROC
PubChem
EF Richter5HT
HTSEF
PubChem TAAR EF
4.5 35
124
30
3.5
10
25
3
8 20
2.5
62 15
1.5
4 10
1
5
0.5
2
0
0
0 0 0.5
0 0.5
1
0 0.5 1
-2
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0 0 0.1 0.2
0
0.1
1
0.90.9
1
0.80.8
0.9
0.70.7
0.8
0.60.6
0.7
0.50.5
0.6
0.40.4
0.5
0.30.3
0.4
0.20.2
0.3
0.10.1
0.2
0 0
0.1
0 0 0.1 0.10.2 0.20.3 0.30.4 0.4
0.5 0.5
0.6 0.6
0.7 0.7
0.8 0.8
0.9
0
FalseFalse
Positive
Positive
RateRate
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
False Positive Rate
True Positive Rate
True Positive Rate
True Positive Rate
1
3
EF
1.5
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
True Positive Rate
3
True Positive Rate
7
Enrichment
3.5
Enrichment
Enrichment
Enrichment
EF
EFChemBL
Richter FS
11.5 1.52 2 2.52.5 3 3 3.5
3.5 44 4.5 55 5.5
5.5 6 6
1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
Fragment
Fragment
GPCR
GPCRScore
Score
Fragment GPCR Score
𝑎𝑐𝑐𝑒𝑝𝑡𝑒𝑑 𝑎𝑐𝑡𝑖𝑣𝑒𝑠
𝐸𝑛𝑟𝑖𝑐ℎ𝑚𝑒𝑛𝑡 =
𝑎𝑙𝑙 𝑎𝑐𝑐𝑒𝑝𝑡𝑒𝑑
𝑎𝑙𝑙 𝑎𝑐𝑡𝑖𝑣𝑒𝑠
𝑎𝑙𝑙
0.9
1
1
1
1
Slide 9
SUMMARY
1. An active set of class-A aminerg-like fragments was extracted from ChemBL GPCR-SARfari
2. Identification of aminerg-characteristic fragment properties
3. Creation of a desirability-function based score (FrAGS – Fragment Aminergic GPCR Score)
4. Validation by using both public and proprietary experimental screening data (PubChem, RG)
FURTHER PLANS
Compilation of an in-house fragment library
Use of FrAGS for the design of a GPCR-targeted fragment library
Supplement of the in-house library by commercially available fragments and by synthesis
Slide 10
Physicochemical property based scoring scheme for
design of an aminerg GPCR-targeted fragment library:
The Fragment-GPCR-Score
„FrAGS”
UGM Budapest 21st of may 2014
Adam A. Kelemen
Hungarian Academy of Sciences
Medicinal Chemistry Research Group
Slide 2
INTRODUCTION:
G-PROTEIN COUPLED RECEPTORS AND THE FRAGMENT-BASED
DRUG DISCOVERY
Fragment-based approach:
Aims of the study:
Biophysical or biochemical screening of small-sized, simple, polar molecules in high-concentrations
Advantages:
1. Better sampling of the chemical space
2. Freedom in optimizing them to lead-molecules with favourable pharmacokinetic properties
Establishment of a scoring scheme supporting the design of GPCR-targeted fragment libraries
Characteristic:
Screening at higher concentrations
Rules for the scoring scheme are derived through:
Fragment-library design:
1. Scaled
to the
fragment-screening
1. data
mining
of an open paradigm
access bioactivity database (ChEMBL)
2. Diversity (fragment chemical space: 107)
using ChemAxon’s Instant Jchem and Knime
G-protein coupled
biology,
2.receptors,
Examination
of localization:
physicochemical descriptors
1. Largest cell-membrane receptor family
(well known from Rule of Three, Rule of Five)
2. Endogenous ligands: neurotransmitters, peptides, hormones
using
ChemAxon’s
for Excel
3. 7TM
structure,
extracellularJChem
N-terminal
4. 85% of GPCRs belongs to class A:
(aminergic-, chemokine-, glycoprotein-hormone-, neuropeptide-)
5. Approx. 40% of marketed small molecule drugs are GPCR-targeted
Stephen P. Andrews, ChemMedChem, 2014, 9, 256-275
Slide 3
DATA COLLECTION AND PREPARATION:
TRAINING SETS
Reference
(inactive)
set
Active
set
GPCRs
Class A
ChEMBL GPCR SARfari
(EBI,
EMBL),(active
Data mining
with Instant
Entire
ChEMBL
set subtracted)
AminergMiner
Jchem and Konstanz Information
947914 entries: molecules,
in vitro-essay data1292344
(binding,entry
functional, ADME), GPCR-target
1. Muscarinic Achetylcholine, ACM[x]
2. Adrenoceptors, α[x], β[x]
Dopamine, DRD[x]
1. Activity data: binding essay, IC50, logIC3.
50, Ki, logKi conversion to pKi
1. Counter ions ofdata
salts stripped 4. Histamine, HRH[x]
2. species-independent
10477 fragments
309962 fragments
5000 random sampling
5. Serotonine, 5HT[x]
2. 8 ions
≤ Nheavy
≤ 22
3. Counter
of salts
stripped
6. Octopamine
4. 8 ≤ Nheavy ≤ 22
7. Trace Amine, TAAR[x]
Rate of aminergic-data
in the active set
derived
from ChEMBL GPCR-SARfari (2370 fragments):
Size-independent
ligand
efficiency:
187 fragments contain𝑝𝐾
at least 4 active data on only not-aminergic class-A receptors
2370 active fragments
𝑆𝐼𝐿𝐸 = [ 𝑖
0,3 ] ≥ 1.95 on at least 4 GPCR-receptors
Remaining 2183 fragments𝑁ℎ𝑒𝑎𝑣𝑦
possess activity data only on aminergic GPCRs
2183 fragments with related activity on aminergic GPCR’s
Slide 4
RED: GPCR fragments
BLUE: inactive reference set
CALCULATION OF PHYSICOCHEMICAL PROPERTIES WITH JCHEM
FOR EXCEL
logP (at pH = 7.4)
Polar Surface Area
0.2
4. Rotatable Bond Count (nROT),
0
0
40
2
3
4
80
100
120
140
logP
Acceptor
Count (HBA)
PSA
60
5.
1
-2
0
0
0
0.6
0.15 0.150
0.1 0.100
Frequency
0.58.
0.2 0.200
Number of Nitrogen Atoms
0.4
9. Number of Oxygen Atoms
0.3
0.050
0.2
0.000
0
1
1
2
2 0.1
3
4
5
3
4
5
Number of Nitrogen Atoms
0 HBA
Base
6
6
Acid
7
7
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
02
31
4 2
5
logD
Rotatable Bond Count
3
6
4 7
5
Number of Oxygen Atoms
Donor Count
0.400
0.350
0.300
Frequency
0.25 0.250
-1 1
BA character
7. Strongest basic pKa
Frequency
Frequency
5
0.8
0.350
0
0.05
0.05
6. Strongest acidic pKa
0.400
0.7
0
160
0.1
0.1
Number of
Nitrogen Atoms
0.9
Acceptor
Count
0.3 0.300
0.05
0.2
0.15
3. Polar Surface Area (PSA)
0.05
20
0.25
Frequency
Frequency
0.1
-1
0.25
2. Octanol:water distribution coefficient
0.2 (logD, pH = 7.4)
0.15
0.15
0
0.3
1. Octanol:water partition coefficient (logP)
0.25
6. Donor Count (HBD)
0.35
logD Bond
(at pHCount
= 7.4)
Rotatable
0.3
Frequency
Frequency
0.5
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
-2
0.250
0.200
0.150
0.100
0.050
0.000
0
Zwitterion
0
1
1
2
Neutral
2
3
4
5
3
4
5
Number
of Oxygen
Atoms
HBD
6
6
7
7
Slide 5
RESULTS I. – SELECTION OF GPCR CHARACTERISTIC DESCRIPTORS
1.
Polar Surface Areas (at pH = 7.4) of the active fragments have their median and mean at lower values,
than the reference set.
2.
logP, Donor Count and Acceptor Count distributions do not show significant difference between the
compared sets, however the lower scores of the logD (calculated at pH = 7.4) showed, that GPCR-like
fragments have mostly basic character.
3.
83% of the active fragments resulted to have basic character, 16% of the fragments were neutral.
Aminergic GPCR-active fragments constitute a set of:
I. small sized
II. rigid molecules,
III. containing few heteroatoms,
IV. that are mostly (~83%) basic nitrogens.
Slide 6
RESULTS II. – DESIRABILITY FUNCTIONS AND SCORING
(FRAGMENT GPCR SCORE)
Rotatable Bond Count - SCORE
The desirability function maps the value of a
10 20 30 40 50 60 70 80 90 100 110 120 130 140
TPSA
0.1
0.05
1
2
3
4
5
6
7 8
pKa
9
10 11 12 13 14
7
8
1
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0.9
0.8
0.7
0
0.6
0.5
0.4
0.3
1
2
3
#O
4
5
6
0.2
Number
of Nitrogen Atoms- SCORE
logD (pH = 7.4) – SCORE
1
0.9
0.8 50 60 70 80 90 100110120130140150
0 10 20 30 40
0.7
0.6
Descriptor
0.5
0.4
0.3
0.2
0.1
0
-3
-2
-1
0
1
2
3
4
5
logD
SCORE
0
FrAGS =
[0; 1].
Desirability Function
Number of Oxygen Atoms- SCORE
SCORE
SCORE
Frequency
0.5
1
0.45
0.9
0.8
0.70.4
0.6
0.35
0.5
0.4
0.30.3
0.2
0.1
0.25
0
0.20
0.15
Distribution of the descriptor
pKa - SCORE
SCORE
0
1
0.9
0.8
0.7
0.6
property
onto a score in the range of
0.5
0.4
0.3
0.2
0.1
0
0
1
2
3
4
5
6
nrot
SCORE
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
SCORE
SCORE
PSA – SCORE
6
0.1
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Descriptor
0
1
2
𝑑𝑒𝑠𝑖𝑟𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛 (𝑝𝑟𝑜𝑝𝑒𝑟𝑡𝑦)
3
#N
4
5
6
Slide 7
VALIDATION – CHEMBL, PUBCHEM, FS, HTS
1. ChEMBL validation set
3. Richter Gedeon experimental validation set
High-throughput and fragment screening data on aminergic GPCR targets
2. PubChem validation sets
class-A targeted HTS confirmed actives with at least 10 μM activity
- screenings on allosteric-bindings
- targeting β-arrestin pathways
- or used as counter-screening were sorted out
Targets represented in HTS campaigns:
5HT1A and 5HT1E with 31 confirmed active fragments
TAAR1 with 169 confirmed active fragments
Slide 8
VALIDATION – CHEMBL, PUBCHEM, FS, HTS
ROC
Richter FS
ROC
ChEMBL
6
2.5
5
2
4
0.5
2
1
0
0
0
0.50
-1
1 1.5
0.5
1 1.52 22.5 2.5 3 3 3.53.5 4 4 4.5
4.5
Fragment
GPCR
Score
Fragment
GPCR
Score
55
5.5 6 6
5.5
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0.2 0.3 0.4 0.5 0.6 0.7 0.8
False Positive Rate
False Positive Rate
0.9
ROC
PubChem
Richter
5HT
HTSROC
1
PubChem
TAAR1 ROC
PubChem
EF Richter5HT
HTSEF
PubChem TAAR EF
4.5 35
124
30
3.5
10
25
3
8 20
2.5
62 15
1.5
4 10
1
5
0.5
2
0
0
0 0 0.5
0 0.5
1
0 0.5 1
-2
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0 0 0.1 0.2
0
0.1
1
0.90.9
1
0.80.8
0.9
0.70.7
0.8
0.60.6
0.7
0.50.5
0.6
0.40.4
0.5
0.30.3
0.4
0.20.2
0.3
0.10.1
0.2
0 0
0.1
0 0 0.1 0.10.2 0.20.3 0.30.4 0.4
0.5 0.5
0.6 0.6
0.7 0.7
0.8 0.8
0.9
0
FalseFalse
Positive
Positive
RateRate
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
False Positive Rate
True Positive Rate
True Positive Rate
True Positive Rate
1
3
EF
1.5
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
True Positive Rate
3
True Positive Rate
7
Enrichment
3.5
Enrichment
Enrichment
Enrichment
EF
EFChemBL
Richter FS
11.5 1.52 2 2.52.5 3 3 3.5
3.5 44 4.5 55 5.5
5.5 6 6
1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
Fragment
Fragment
GPCR
GPCRScore
Score
Fragment GPCR Score
𝑎𝑐𝑐𝑒𝑝𝑡𝑒𝑑 𝑎𝑐𝑡𝑖𝑣𝑒𝑠
𝐸𝑛𝑟𝑖𝑐ℎ𝑚𝑒𝑛𝑡 =
𝑎𝑙𝑙 𝑎𝑐𝑐𝑒𝑝𝑡𝑒𝑑
𝑎𝑙𝑙 𝑎𝑐𝑡𝑖𝑣𝑒𝑠
𝑎𝑙𝑙
0.9
1
1
1
1
Slide 9
SUMMARY
1. An active set of class-A aminerg-like fragments was extracted from ChemBL GPCR-SARfari
2. Identification of aminerg-characteristic fragment properties
3. Creation of a desirability-function based score (FrAGS – Fragment Aminergic GPCR Score)
4. Validation by using both public and proprietary experimental screening data (PubChem, RG)
FURTHER PLANS
Compilation of an in-house fragment library
Use of FrAGS for the design of a GPCR-targeted fragment library
Supplement of the in-house library by commercially available fragments and by synthesis
Slide 10