DOI: 10.1038/NCHEM.1297 Multidimensional Steric Parameters ... · Matrix form and Y were stepwise...
Transcript of DOI: 10.1038/NCHEM.1297 Multidimensional Steric Parameters ... · Matrix form and Y were stepwise...
NATURE CHEMISTRY | www.nature.com/naturechemistry 1
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
Multidimensional Steric Parameters in the Analysis of Asymmetric Catalytic Reactions
Supporting Information
Kaid C. Harper, Elizabeth N. Bess, Matthew S. Sigman
Table of Contents
General Information S2
Steric Parameters S3
NHK Allylation of Benzaldehyde Data S5
Model Development for the Allylation of Benzaldehyde S6
NHK Allylation of Acetophenone Data S10
Model Development for the Allylation of Acetophenone S11
The Desymmetrization of Bisphenol Data S15
Model Determination for the Desymmetrization of Bisphenols S16
NHK Allylation of Benzaldehyde Data Using the Oxazoline
Proline Ligand Library S20
Attempts at Model Determination for the NHK Allylation of
Benzaldehyde Using the Oxazoline Proline Ligand Library S21
Principal Component Analysis for the Allylation of Benzaldehyde
Using the Oxazoline Proline Ligand Library S25
NHK Propargylation of Acetophenone Data S26
Model Determination for the Propargylation of Acetophenone S27
Principal Component Analysis for the Propargylation of Acetophenone S32
References S33
Multidimensional Steric Parameters in the Analysis of Asymmetric Catalytic Reactions
Supporting Information
Kaid C. Harper, Elizabeth N. Bess, Matthew S. Sigman
Table of Contents
General Information S2
Steric Parameters S3
NHK Allylation of Benzaldehyde Data S5
Model Development for the Allylation of Benzaldehyde S6
NHK Allylation of Acetophenone Data S10
Model Development for the Allylation of Acetophenone S11
The Desymmetrization of Bisphenol Data S15
Model Determination for the Desymmetrization of Bisphenols S16
NHK Allylation of Benzaldehyde Data Using the Oxazoline
Proline Ligand Library S20
Attempts at Model Determination for the NHK Allylation of
Benzaldehyde Using the Oxazoline Proline Ligand Library S21
Principal Component Analysis for the Allylation of Benzaldehyde
Using the Oxazoline Proline Ligand Library S25
NHK Propargylation of Acetophenone Data S26
Model Determination for the Propargylation of Acetophenone S27
Principal Component Analysis for the Propargylation of Acetophenone S32
References S33
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 2
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
General Information
All data used in this report were previously reported, and the reader is directed to the primary publications for information regarding ligand characterization, ligand synthesis, reaction conditions and other reaction outcomes. Information regarding the Nozaki‐Hiyama‐Kishi allylation of benzaldehyde and acetophenone can be found in SI references 1 &2.1,2 Information regarding the desymmetrization of bis‐phenols can be found in SI references 3 & 4.3,4 Information regarding the 3D library for the Nozaki‐Hiyama‐Kishi reactions can be found in SI reference 5.5 Information regarding the propargylation of ketones can be found in SI reference 6.6
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 3
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
Steric Parameters
Table S1. Literature values for the substituents used in this report.
Substituent Interference (kcal/mol)
A (kcal/mol) Charton Molar
Refractivity Sterimol B1 Sterimol B5
Sterimol L
H 0.956 0 0 0.1 1 1 1.52 Me 9.7275 1.7 0.52 0.57 1.52 2.04 2.87 Et 1.8 0.56 1.03 1.52 3.17 4.11 Ph 7.911 2 0.57 2.54 1.71 3.11 6.28 i‐Pr 12.57 2.1 0.76 1.5 1.9 3.17 4.11 Cy 2.2 0.87 2.67 1.91 3.49 6.17 t‐Bu 18.308 4.4 1.24 1.97 2.6 3.17 4.11
CH2t‐Bu 2 1.34 1.52 4.18 4.89 meIpr 0.98 1.52 4.45 4.92 CHEt2 1.51 2.13 4.01 4.72 1‐Ad 1.33 3.16 3.49 6.17 Cet3 2.38 2.94 4.18 4.92 CHPr2 1.54 1.9 4.54 6.17 Bn 0.7 1.52 6.02 4.62
Mean Value 9.89 2.03 0.73 1.48 1.86 3.57 4.68
Std. Deviation 6.36 1.19 0.43 0.98 0.40 1.19 1.35
Table S1 lists the literature values for the steric parameters found in this report. More extensive lists of parameters can be obtained from References 7‐16.7‐16 For our purposes of comparison, the parameters were normalized according to the definition:
XN = (X – Xave) / sx
Where XN is the normalized value, X is the literature value, Xave is the mean value for the range of X, and sx is the standard deviation for the range in X. Using this equation, the Tables S2 and S3 were obtained, which were reported in the text.
Table S2. Normalization scores for small substituents in several parameters.
Substituent Interference (kcal/mol) A (kcal/mol) Charton Molar
Refractivity H ‐1.405 ‐1.523 ‐1.710 ‐1.417
Me ‐0.026 ‐0.360 ‐0.496 ‐0.935
Et ‐0.338 ‐0.403 ‐0.464
Ph ‐0.312 ‐0.601 ‐0.379 1.083
i‐Pr 0.421 0.016 0.064 0.018
Cy ‐0.338 0.321 1.216
t‐Bu 1.323 0.777 1.185 0.499
CH2t‐Bu ‐0.338 1.418
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 4
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
Table S3. Comparison of Normalized Charton parameters and Sterimol B1 parameters.
Substituent Charton Sterimol B1
H ‐1.710 ‐2.143
Me ‐0.496 ‐0.847
Et ‐0.403 ‐0.847
Ph ‐0.379 ‐0.374
i‐Pr 0.064 0.100
Cy 0.321 0.125
t‐Bu 1.185 1.844
CH2t‐Bu 1.418 ‐0.847
meIpr 0.578 ‐0.847
CHEt2 1.815 0.673
1‐Ad 1.395 3.239
Cet3 3.847 2.691
CHPr2 1.885 0.100
Bn ‐0.076 ‐0.847
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 5
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
NHK Allylation of Benzaldehyde Data
Figure S1. The NHK allylation of benzaldehyde using Oxazoline‐Proline ligand series 1.
Table S4. Data gathered in the NHK allylation of benzaldehyde using conditions in Fig. S1.
X Group Major
Enantiomer Minor
Enantiomer ΔΔG‡ Me 60 40 0.245 Et 66 34 0.382 i‐Pr 82 18 0.913 tBu 96 4 1.854
CH(Pr)2 73 27 0.598 CEt3 92 8 1.411
CH(iPr)2 77 23 0.730 1‐Ad 96 4 1.854
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 6
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
Model De
Tha version The raw dand meas
Table S5.
Matrices XNext, the the prom
Figure S2
Fig. S2 shothe upper
evelopment f
he benzaldehthat containsdata was comsured respons
Matrix form
X and Y werestepwise solvpt window sh
. The stepwis
ows the coeffr right box an
for the Allylat
hyde Sterimols the statisticspiled into nuse, respective
atting of the
Y Group Me Et i‐Pr tBu
CH(Pr)2 CEt3
CH(iPr)2 1‐Ad
input into thver was initiahown in Fig. X
se regression
ficients for ead also indicat
tion of Benza
model was ds toolbox is remeric matriceely, as shown
benzaldehyd
MaB1 1.521.521.92.61.9
2.942.083.16
he Matlab™ soted using theXX.
prompt.
ach term (X1, tes whether e
aldehyde
derived using equired to aces X and Y accin Table S5.
e data.
atrix X B5 L2.04 23.17 43.17 43.17 44.54 64.18 44.19 43.49 6
oftware usinge command “s
X2, and X3 coeach term is (
Matlab™ sofccess the stepcording to lig
MatL ΔΔ2.87 0.24.11 0.34.11 0.94.11 1.86.17 0.54.92 1.44.12 0.76.17 1.8g X = [Matrix Xstepwise(X,Y)
orrespond to(blue) or is no
ftware. The spwise linear regand substitue
trix Y ΔG‡ 2454 3818 9135 8543 5977 4106 7300 8543 X] and Y = [M)”. This comm
o B1, B5, and ot (red) includ
student versioegression tooent paramete
Matrix Y] notatmand produce
L, repsectivelded in the cur
on or ol. ers
tions. ed
y) in rrent
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 7
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
model. Fivalue, as wright box.box is a hclicking ondisplay sh
Figure S3
In Fig. S3,model. Thcoefficiensubjectedthe L termindicator)
Figure S4
ig. S2 also showell as select With no teristory of the vn each term ihown in Fig. S
. Initial iterat
note that all he “Model Hisnts are estimad to t‐ and p‐tm is insignifica). Removal of
. Second iter
ows that no tted relevant sms included ivalues of the n the upper r3.
tion in the be
of the termsstory” now haated for the Bests to measuant and shoulf this term lea
ation in the b
erms are inclstatistics, for tin the model,root‐mean‐sright hand bo
nzaldehyde r
are shown inas 4 points foB1, B5 and L teure their signld be eliminatads to the mo
benzaldehyde
uded into thethe model giv these statistquare error (ox, each term
regression.
n blue, which or the separaterms in the uificance in thted from the odel shown in
e regression.
e model. Theven by the blutics are meanRMSE) of eacis incorporat
means all terte inclusion opper right hahe model at hmodel (as prn Fig. S4.
e center box gue coefficientingless in Fig.ch model calcted into the m
rms are incluf each of the and box. Thesand. These sompted by th
gives the intets in the uppe. S2. The botculated. By model to give
ded in the prthree terms.se values are statistics showhe “Next Step
rcept er tom
the
esent The
w that p”
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 8
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
Fig. S4 shomodel pro
Figure S5
Fig. S5 sho
Using this
ows the modompts us to e
. Final iterati
ows the mod
s model, each
el that excludeliminate the
on for the be
el that contai
h of the synth
des the L termB5 term.
enzaldehyde r
ins only the B
ΔΔG‡ = ‐1.
etic ligands’ e
m. Again, ana
regression.
B1 term and a
068 + 0.938 B
enantioselect
alysis of the st
an intercept v
B1 (1)
tivities were p
tatistical mea
value shown i
predicted (Ta
asures for this
n Eq. 1.
ble S6).
s
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 9
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
Table S6. Comparison of experimentally measured and predicted (from Eq. 1) enantioselectivities for the NHK allylation of benzaldehyde.
Measured ΔΔG‡
Predicted ΔΔG‡
0.245 0.359 0.382 0.359 0.913 0.715 1.854 1.372 0.598 0.715 1.411 1.691 0.730 0.884 1.854 1.897
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 10
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
NHK Allylation of Acetophenone Data
Figure S6. Standard conditions for the NHK allylation of acetophenone.
Table S7. Raw data for the allylation of acetophenone.
Y Group Enantiomer 1 Enantiomer 2 ΔΔG‡ Me 25 75 ‐0.650 Et 24 76 ‐0.680 i‐Pr 38 62 ‐0.300 tBu 69 31 0.470
CH(Pr)2 38 62 ‐0.300 CEt3 62 38 0.280
CH(iPr)2 55 45 0.120 1‐Ad 72 28 0.566
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 11
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
Model Development for the Allylation of Acetophenone
The acetophenone Sterimol model was derived using Matlab™ software. The student version, or a version which contains the statistics toolbox, is required to access the stepwise linear regression tool. The raw data was compiled into numeric matrices X and Y according to ligand substituent parameters and measured response, respectively, as shown in Table S8.
Table S8. Matrix format for the allylation of acetophenone data.
Matrix X Matrix Y Y Group B1 B5 L ΔΔG‡ Me 1.52 2.04 2.87 ‐0.650 Et 1.52 3.17 4.11 ‐0.680 i‐Pr 1.9 3.17 4.11 ‐0.300 tBu 2.6 3.17 4.11 0.470
CH(Pr)2 1.9 4.54 6.17 ‐0.300 CEt3 2.94 4.18 4.92 0.280
CH(iPr)2 2.08 4.19 4.12 0.120 1‐Ad 3.16 3.49 6.17 0.566
Matrices X and Y were input into the Matlab™ software using X = [Matrix X] and Y = [Matrix Y] notations. Then the stepwise solver was initiated using the command “stepwise(X,Y)”. This command prodcued the prompt window shown in Fig. S7.
Figure S7. Initial regression command prompt.
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 12
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
This promrespectiveincluded igives the coefficienin Fig. S7. model calthe mode
Figure S8
mpt shows theely) in the upin the currentintercept valunts in the upp The bottomlculated. By cel to give the d
. Initial aceto
e coefficients per right box t model. Fig. ue as well as er right box. W box is a histolicking on eacdisplay shown
ophenone iter
for each termand also indiS7 shows thaselected releWith no termory of the valch term in then in Fig. S8.
ration.
m (X1, X2, andicates whetheat no terms avant statistic
ms included inues of the rooe upper right
d X3 correspoer each term re included ins for the modn the model, tot‐mean‐squahand box, ea
ond to B1, B5is (blue) or isnto the modedel given by tthese statisticare error (RMach term is in
, and L, s not (red) el. The centehe blue cs are meaninMSE) of each corporated in
r box
ngless
nto
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 13
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
In Fig. S3,present mterms. Thvalues areshows tha“Next Ste
Figure S9
note that all model. The mohe coefficiente also subjectat the B5 termp” indicator).
. Second ace
the terms arodel history nts are estimatted to t‐ and pm is insignifica. Removal of
tophenone it
e shown in bnow has 4 poited for the B1p‐tests to meant and shouthis term lea
teration.
lue, which meints for the se1, B5 and L teasure their sild be eliminatads to the mo
eans that all teparate inclusrms in the upignificance inted from the odel shown in
terms are incsion of each opper right han the model atmodel (as pr Fig. S9.
cluded in the of the three nd box. Theset hand. This rompted by th
e
he
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 14
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
Fig. S9 shoreflects ththat the L
Figure S10
Fig. S10 gmodel staterm, whi
Using Eq. the measu
Table S9. acetophe
ows the B5 tehe changes inL term is insig
0. Final iterat
ives the final atistics with eich produce E
2, the enantiured values (T
Experimentanone.
erm in red ind coefficient vnificant and s
tion in the m
model after txclusion of thEq. 2.
ioselectivitiesTable S9).
ally measured
dicating that ivalues that acshould be elim
odeling of the
the exclusionhe L term. Th
ΔΔG‡ = ‐1
s for each of t
d values comp
MeasureΔΔG‡
‐0.650‐0.680‐0.3000.470‐0.3000.2800.1200.566
it has been excompany excminated from
e acetopheno
n of the L termhe only remai
1.67 + 0.73 B
the synthetic
pared to valu
ed PredΔΔ‐0.5‐0.5‐0.20.2‐0.20.4‐0.10.6
xcluded fromclusion of them the model.
one data.
m. There is signing terms ar
1 (2)
ligands were
es calculated
icted ΔG‡ 560 560 283 228 283 476 152 637
m the model ate B5 term. Fig
gnificant impre the offset t
e predicted an
d from Eq. 2 in
t hand. It alsg. S9 also sho
provement in term and the
nd compared
n the allylatio
o ws
the e B1
with
on of
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 15
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
The Desymmetrization of Bisphenol Data
Figure S11. Standard conditions for the desymmetrization reaction.
Table S10. Desymmetrization data with replicate runs.
R Group Major
Entantiomer Minor
Enantiomer ΔΔG‡ Me 76 24 0.543 Et 78 22 0.596 Ph 75 25 0.517 Bn 71 29 0.421 iPr 87 13 0.895 tBu 97.5 2.5 1.724 Cy 83 17 0.746
CH2tBu 77.5 22.5 0.582 CHEt2 92 8 1.150 CH2iPr 67 33 0.333 CHPh2 84 16 0.780 1‐Ad 97.5 2.5 1.724 Me 76 24 0.543 Et 78.5 21.5 0.610 Ph 74.5 25.5 0.505 Bn 72.5 27.5 0.456 iPr 86.5 13.5 0.874 tBu 97.5 2.5 1.724 Cy 81.5 18.5 0.698
CH2tBu 77.5 22.5 0.582 CHEt2 91 9 1.089 CH2iPr 68.5 31.5 0.366 CHPh2 86 14 0.854 1‐Ad 97.3 2.7 1.687
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 16
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
Model Determination for the Desymmetrization of Bisphenols
The desymmetrization Sterimol model was derived using Matlab™ software. The student version, or a version which contains the statistics toolbox, is required to access the stepwise linear regression tool. The raw data was compiled into numeric matrices X and Y according to ligand substituent parameters and measured response, respectively, as shown in Table S11.
Table S11. Matrix format for the desymmetrization data.
Matrix X Matrix Y
R Group B1 B5 L ΔΔG‡ Me 1.52 2.04 2.87 0.543 Et 1.52 3.17 4.11 0.596 Ph 1.71 3.11 6.28 0.517 Bn 1.52 6.02 4.62 0.421 iPr 1.9 3.17 4.11 0.895 tBu 2.6 3.17 4.11 1.724 Cy 1.91 3.49 6.17 0.746
CH2tBu 1.52 4.18 4.89 0.582 CHEt2 2.13 4.01 4.72 1.150 CH2iPr 1.52 4.45 4.92 0.333 CHPh2 2.01 6.02 5.15 0.780 1‐Ad 3.16 3.49 6.17 1.724 Me 1.52 2.04 2.87 0.543 Et 1.52 3.17 4.11 0.610 Ph 1.71 3.11 6.28 0.505 Bn 1.52 6.02 4.62 0.456 iPr 1.9 3.17 4.11 0.874 tBu 2.6 3.17 4.11 1.724 Cy 1.91 3.49 6.17 0.698
CH2tBu 1.52 4.18 4.89 0.582 CHEt2 2.13 4.01 4.72 1.089 CH2iPr 1.52 4.45 4.92 0.366 CHPh2 2.01 6.02 5.15 0.854 1‐Ad 3.16 3.49 6.17 1.687
Matrices X and Y were input into the Matlab™ software using X = [Matrix X] and Y = [Matrix Y] notations. The stepwise solver was initiated using the command “stepwise(X,Y)”. This command produced the prompt window shown in Fig. S12.
Figure S12. Initial stepwise regression prompt.
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 17
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
This promrespectivein the curintercept upper righ
With no thistory ofterm in thFig. S13.
Figure S13
mpt shows theely) in the uprent model. value, as welht box.
erms includef the values ofhe upper right
3. Initial inclu
e coefficients per right box Fig. S12 showll as selected
d in the modef the root‐met hand box, e
usive model f
for each termand also indi
ws that no terrelevant stat
el, these statean‐square erach term is in
for the desym
m (X1, X2, andicates whetherms are includistics for the
istics are mearror (RMSE) oncorporated i
mmetrization
d X3 correspoer the term isded in the momodel given
aningless in Fof each modeinto the mod
reaction.
ond to B1, B5s (blue) or is nodel. The cenby the blue c
Fig. S12. The l calculated. Bel to give the
, and L, not (red) inclunter box givescoefficients in
bottom box iBy clicking one display show
uded s the n the
s a n each wn in
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 18
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
In Fig. S14present mterms. Thvalues arestatistics sby the “N
Figure S14
4, note that amodel. The mohe coefficiente subjected toshow that theext Step” ind
4. First and f
ll the terms aodel history nts are estimato t‐ and p‐tese B5 term is iicator). Rem
inal iteration
are shown in bnow has 4 poited for the B1ts to measurensignificant aoval of this te
of the desym
blue, which mints for the se1, B5 and L tee their signifiand should beerm leads to t
mmetrization
means that aleparate inclusrms in the upcance in the e eliminated fthe model sh
model.
l terms are insion of each opper right hanmodel at hanfrom the modhown in Fig. S
ncluded in theof the three nd box. Thesend. These del (as promp14.
e
e
pted
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 19
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
Exclusion of the B5 term leads to the model with the highest degree of statistical significance shown in Eq. 3.
ΔΔG‡ = ‐0.417 + 0.928 B1 – 0.110 L (3)
Using this model, predictions were calculated for all the synthetic substrates, generating the data in Table S12.
Table S12. Experimentally measured values compared to those given by Eq. 3 for the desymmetrization reaction.
Measured ΔΔG‡
Predicted ΔΔG‡
0.543 0.680 0.596 0.544 0.517 0.483 0.421 0.488 0.895 0.897 1.724 1.547 0.746 0.681 0.582 0.458 1.150 1.044 0.333 0.455 0.780 0.885 1.724 1.841 0.543 0.680 0.610 0.544 0.505 0.483 0.456 0.488 0.874 0.897 1.724 1.547 0.698 0.681 0.582 0.458 1.089 1.044 0.366 0.455 0.854 0.885 1.687 1.841
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 20
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
NHK Allylation of Benzaldehyde Data Using the Oxazoline Proline Ligand Library ***
Figure S15. Standard conditions for the allylation of benzaldehyde.
For the complete data set, refer to previously published work.6
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 21
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
Attempts at Model Determination for the NHK Allylation of Benzaldehyde Using the Oxazoline Proline Ligand Library
Because the data analyzed herein was generated prior to our conception of applying modeling techniques to determine an optimized ligand, only those data points representing an evenly spaced data spread were included in the matrices for model determination. Table S13 presents all experimental ΔΔG‡values, averaged over two runs, and Sterimol parameters for the substituent combinations evaluated. Substituent combinations highlighted in yellow are those used for model determination. Table S13. Full data set for NHK allylation of benzaldehyde.
Substituent Combination a,b XB1 XB5 XL YB1 YB5 YL
Measured ΔΔG‡
HM 1 1 2.06 1.52 2.04 2.87 0.0000 HE 1 1 2.06 1.52 3.17 4.11 0.3004 HI 1 1 2.06 1.9 3.17 4.11 0.2541 HB 1 1 2.06 2.6 3.17 4.11 0.9410 HD 1 1 2.06 1.9 4.54 6.17 0.6464 HT 1 1 2.06 2.94 4.18 4.92 0.1640 MM 1.52 2.04 2.87 1.52 2.04 2.87 0.1308 ME 1.52 2.04 2.87 1.52 3.17 4.11 0.2427 MI 1.52 2.04 2.87 1.9 3.17 4.11 0.3239 MB 1.52 2.04 2.87 2.6 3.17 4.11 0.8602 MD 1.52 2.04 2.87 1.9 4.54 6.17 1.0808 MT 1.52 2.04 2.87 2.94 4.18 4.92 0.3598 EM 1.52 3.17 4.11 1.52 2.04 2.87 0.2427 EE 1.52 3.17 4.11 1.52 3.17 4.11 0.5816 EI 1.52 3.17 4.11 1.9 3.17 4.11 0.8796 EB 1.52 3.17 4.11 2.6 3.17 4.11 0.8226 ED 1.52 3.17 4.11 1.9 4.54 6.17 1.1342 ET 1.52 3.17 4.11 2.94 4.18 4.92 0.1308 IM 1.9 3.17 4.11 1.52 2.04 2.87 0.2771 IE 1.9 3.17 4.11 1.52 3.17 4.11 0.8796 II 1.9 3.17 4.11 1.9 3.17 4.11 0.9626 IB 1.9 3.17 4.11 2.6 3.17 4.11 1.2891 ID 1.9 3.17 4.11 1.9 4.54 6.17 0.6164 IT 1.9 3.17 4.11 2.94 4.18 4.92 0.0217 BM 2.6 3.17 4.11 1.52 2.04 2.87 0.2290 BE 2.6 3.17 4.11 1.52 3.17 4.11 0.1974 BI 2.6 3.17 4.11 1.9 3.17 4.11 0.3299 BB 2.6 3.17 4.11 2.6 3.17 4.11 0.3358 BD 2.6 3.17 4.11 1.9 4.54 6.17 0.4214
a First letter is X substituent. Second letter is Y substituent. b Substituent codes: H= ‐H, M= ‐CH3, E= ‐CH2CH3, I= ‐CH(CH3)2, B= ‐C(CH3)3, D= ‐CH(C3H7)2, T= ‐C(C2H5)3 Selection of substituent combinations for inclusion in the model data set for model determination was based upon variation in B5. This selection criterion was chosen due to B5 exhibiting the largest spread of
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 22
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
values. When B5 values were similar, the decision for inclusion of a substituent combination in the modeling data set was based upon the degree of spread in B1 values, as L values increase in a fashion similar to the increase seen in B5. Inclusion of all possible crossterms resulted in matrices X and Y (Table S14).
Table S14. Matrix format for the allylation data.
Matrix X Matrix Y
XB1 XB5 XL YB1 YB5 YL XB1YB1 XB1YB5 XB1YL XB5YB1 XB5YB5 XB5YL XLYB1 XLYB5 XLYL ΔΔG‡
1.00 1.00 2.06 1.52 2.04 2.87 1.52 2.04 2.87 1.52 2.04 2.87 3.13 4.20 5.91 0.0000
1.00 1.00 2.06 1.90 3.17 4.11 1.90 3.17 4.11 1.90 3.17 4.11 3.91 6.53 8.47 0.2541
1.00 1.00 2.06 2.60 3.17 4.11 2.60 3.17 4.11 2.60 3.17 4.11 5.36 6.53 8.47 0.9410
1.00 1.00 2.06 2.94 4.18 4.92 2.94 4.18 4.92 2.94 4.18 4.92 6.06 8.61 10.14 0.1640
1.52 2.04 2.87 1.52 2.04 2.87 2.31 3.10 4.36 3.10 4.16 5.85 4.36 5.85 8.24 0.1308
1.52 2.04 2.87 1.90 3.17 4.11 2.89 4.82 6.25 3.88 6.47 8.38 5.45 9.10 11.80 0.3239
1.52 2.04 2.87 2.60 3.17 4.11 3.95 4.82 6.25 5.30 6.47 8.38 7.46 9.10 11.80 0.8602
1.52 2.04 2.87 2.94 4.18 4.92 4.47 6.35 7.48 6.00 8.53 10.04 8.44 12.00 14.12 0.3598
1.90 3.17 4.11 1.52 2.04 2.87 2.89 3.88 5.45 4.82 6.47 9.10 6.25 8.38 11.80 0.2771
1.90 3.17 4.11 1.90 3.17 4.11 3.61 6.02 7.81 6.02 10.05 13.03 7.81 13.03 16.89 0.9626
1.90 3.17 4.11 2.60 3.17 4.11 4.94 6.02 7.81 8.24 10.05 13.03 10.69 13.03 16.89 1.2891
1.90 3.17 4.11 2.94 4.18 4.92 5.59 7.94 9.35 9.32 13.25 15.60 12.08 17.18 20.22 0.0217
2.60 3.17 4.11 1.52 2.04 2.87 3.95 5.30 7.46 4.82 6.47 9.10 6.25 8.38 11.80 0.2290
2.60 3.17 4.11 1.90 3.17 4.11 4.94 8.24 10.69 6.02 10.05 13.03 7.81 13.03 16.89 0.3299
2.60 3.17 4.11 2.60 3.17 4.11 6.76 8.24 10.69 8.24 10.05 13.03 10.69 13.03 16.89 0.3358
Matrices X and Y were input into the Matlab™ software using X = [Matrix X] and Y = [Matrix Y] notations. Then, the stepwise solver was initiated using the command “stepwise(X,Y)”. This command produced the prompt window shown in Fig. S16.
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 23
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
Figure S16
This promX14, and XXLYB1, XLYnot (red) center bocoefficienmeaningleof each m
6. Initial Regr
mpt shows theX15 correspoYB5, and XLYL, included in thox gives the innts in the uppess in Fig. S16
model calculat
ression Comm
e coefficients ond to terms Xrespectively) he current montercept valueer right box. 6. The bottomted.
mand Prompt.
for each termXB1, XB5, XL, YBin the upper odel. Fig. S16e and other reWith no termm box is a his
.
m (X1, X2, X3,1, YB5, YL, XB1Yright box and6 shows that elevant statisms included intory of the va
, X4, X5, X6, XYB1, XB1, YB5, Xd also indicatno terms are stics for the mn the model, alues of the r
X7, X8, X9, X1B1, YL, XB5YB1, es whether e included in tmodel given bthese statistioot‐mean‐sq
0, X11, X12, XXB5, YB5, XB5YLeach is (blue) the model. Thby the blue cs are uare error (R
X13, L, or is he
RMSE)
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 24
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
Fitting a linear relationship to this data set proved problematic. The data matrix (Matrix X) contains 15 variables and 15 data points. Thus, the starting model for backward elimination stepwise regression (which includes all terms in the initial model) suffered from over‐fitting. Alternatively, forward selection stepwise regression was considered. However, as seen in Fig. S16, there were no terms that could be included in the model to increase the model fit.
Attempts at modeling via partial least squares (PLS) regression highlighted the problematic covariance amongst the Sterimol parameters (Table 3), as well as additional confounding aspects of the data set, which remain under investigation.
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 25
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
Principal Component Analysis for the Allylation of Benzaldehyde
The principal component analysis was performed in Matlab™ from Matrix X (X) using the commands:
[coefs, scores, variances]=princomp(X) percent_explained = 100*variances/sum(variances)
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 26
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
NHK Propargylation of Acetophenone Data
Figure S17. Standard conditions for the QuinPro ligand library evaluation in the NHK propargylation reaction.
Table S15. Data, with replicate runs, obtained from the ligand library evaluation.
Electronic Group (E)
Steric Group (S) Peak 1 Peak 2 ΔΔG‡
CF3 Me 44.6 55.4 ‐0.1270 CF3 tBu 50.6 49.4 0.0141 CF3 CEt3 60 40 0.2375 H Me 53 47 0.0704 H tBu 86.8 13.2 1.1034 H CEt3 76.4 23.6 0.6882
OMe Me 60.2 39.8 0.2424 OMe tBu 90.8 9.2 1.3413 OMe CEt3 79.9 20.1 0.8085 CF3 Me 47 53 ‐0.0704 CF3 tBu 62 38 0.2868 CF3 CEt3 62 38 0.2868 H Me 55 45 0.1176 H tBu 87.5 12.5 1.1401 H CEt3 78.8 21.2 0.7692
OMe Me 60 40 0.2375 OMe tBu 92 8 1.4309 OMe CEt3 82 18 0.8884
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 27
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
Model Determination for the Propargylation of Acetophenone
The propargylation Sterimol model was derived using Matlab™ software. The student version, or a version which contains the statistics toolbox, is required to access the stepwise linear regression tool. The raw data was compiled into numeric matrices X and Y according to ligand substituent parameters and measured response, respectively, as shown in Table S16. Matrix X differs from previous like matrices because the data set contains variation in steric and electronic parameters. Thus, Matrix X includes crossterms between the electronic parameter and each of the Sterimol parameters. The crossterms are derived by multiplying the respective parameters for a given ligand.
Table S16. Matrix format for the raw propargylation data.
Matrix X Matrix Y
B1 B5 L σ σB1 σB5 σL ΔΔG‡ 1.52 2.04 2.87 0.54 0.8208 1.1016 1.5498 ‐0.127042.6 3.17 4.11 0.54 1.404 1.7118 2.2194 0.014062
2.94 4.18 4.92 0.54 1.5876 2.2572 2.6568 0.237551.52 2.04 2.87 0 0 0 0 0.0703892.6 3.17 4.11 0 0 0 0 1.103422
2.94 4.18 4.92 0 0 0 0 0.6882431.52 2.04 2.87 ‐0.27 ‐0.4104 ‐0.5508 ‐0.7749 0.2424362.6 3.17 4.11 ‐0.27 ‐0.702 ‐0.8559 ‐1.1097 1.341323
2.94 4.18 4.92 ‐0.27 ‐0.7938 ‐1.1286 ‐1.3284 0.8085331.52 2.04 2.87 0.54 0.8208 1.1016 1.5498 ‐0.070392.6 3.17 4.11 0.54 1.404 1.7118 2.2194 0.286812
2.94 4.18 4.92 0.54 1.5876 2.2572 2.6568 0.2868121.52 2.04 2.87 0 0 0 0 0.1175672.6 3.17 4.11 0 0 0 0 1.14005
2.94 4.18 4.92 0 0 0 0 0.7691961.52 2.04 2.87 ‐0.27 ‐0.4104 ‐0.5508 ‐0.7749 0.237552.6 3.17 4.11 ‐0.27 ‐0.702 ‐0.8559 ‐1.1097 1.430898
2.94 4.18 4.92 ‐0.27 ‐0.7938 ‐1.1286 ‐1.3284 0.888383Matrices X and Y were input into the Matlab™ software using X = [Matrix X] and Y = [Matrix Y] notations. The stepwise solver was initiated using the command “stepwise(X,Y)”. This command produced the prompt window shown in Fig. S18.
Figure S18. Initial regression command prompt.
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 28
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
This promσ, σB1, σBnot (red) center boblue coeffmeaningleof each mincorpora
Figure S19
mpt shows theB5, and σL, reincluded in thox gives the inficients in theess in Fig. S18
model calculatated into the m
9. Initial inco
e coefficients espectively) inhe current montercept valuee upper right 8. The bottomted. By clickinmodel to give
orporation of
for each termn the upper riodel. Fig. S18e as well as sebox. With nom box is a hisng on each tere the display s
all variables i
m (X1, X2, X3,ght box and i8 shows that elected releva terms includtory of the varm in the uppshown in Fig.
in the regress
, X4, X5, X6, aindicates wheno terms are ant statistics ded in the moalues of the rper right hand S19.
sion.
and X7 corresether each te included intofor the modeodel, these staoot‐mean‐sqd box, each te
pond to B1, Brm is (blue) oo the model. el given by theatistics are uare error (Rerm is
B5, L, or is The e
RMSE)
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 29
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
In Fig. S19present mThe coeffbox. Theshand. Thiprompted
Figure S20
Fig. S20 i
Figure S2
9, note that amodel. The moicients are esse values are is shows thatd by the “Nex
0. First iterat
ndicates that
1. Second ite
ll the terms aodel history ntimated for talso subjecte the X7 term xt Step” indica
tion in the pro
t the X4 term
eration in the
are shown in bnow has 8 poihe B1, B5, L aed to t‐ and p(σL) is insigniator). Remov
opargylation
(σ) should be
propargylati
blue, which mints for the seand σ terms a‐tests to meaificant and shval of this term
regression.
e eliminated,
on regression
means that aleparate inclusand their crosasure their sighould be elimm leads to the
generating th
n.
l terms are insion of each ossterms in thegnificance in tinated from te model show
he model sho
ncluded in theof the eight tee upper right the model at the model (aswn in Fig. S20
own in Fig. S2
e erms. hand
s 0.
1.
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 30
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
Although further simsignificant
Figure S22
This mode
Eq. 4 wasmeasure
this model apmplified throt model show
2. Final iterat
el represents
ΔΔG
s used to pred values (Ta
ppears adequugh eliminatiwn in Fig. S22.
tion of the pr
the most sim
G‡ = ‐0.696 +
edict values ble S17).
uate, throughion of the L te.
ropargylation
mplified statis
+ 1.83 B1 – 0
for all synth
inspection iterm. Elimina
model.
tically signific
0.962 B5 – 2.7
hetic ligands,
t was determation of the L
cant model an
705 EB1 + 1.
, which were
ined that theterm gives an
nd is shown i
736 EB5 (4)
e then comp
model couldn equally
n Eq. 4.
pared to the
d be
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 31
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
Table S17. Experimentally measured values compared to values predicted from Eq. 4.
Measured ΔΔG‡
Predicted ΔΔG‡
‐0.127 ‐0.178 0.014 0.198 0.238 0.300 0.070 0.130 1.103 1.024 0.688 0.676 0.242 0.284 1.341 1.437 0.809 0.864 ‐0.070 ‐0.178 0.287 0.198 0.287 0.300 0.118 0.130 1.140 1.024 0.769 0.676 0.238 0.284 1.431 1.437 0.888 0.864
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 32
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
Principal Component Analysis for the Propargylation of Acetophenone
The principal component analysis plot of this data was generated in Matlab™ from Matrix X (X) using the commands:
[coefs, scores, variances]=princomp(X) percent_explained = 100*variances/sum(variances)
© 2012 Macmillan Publishers Limited. All rights reserved.
NATURE CHEMISTRY | www.nature.com/naturechemistry 33
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.1297
References
(1) Miller, J. J.; Sigman, M. S. Angew. Chem., Int. Ed. 2008, 47, 771. (2) Sigman, M. S.; Miller, J. J. J. Org. Chem. 2009, 74, 7633. (3) Lewis, C. A.; Gustafson, J. L.; Chiu, A.; Balsells, J.; Pollard, D.; Murry, J.; Reamer, R. A.; Hansen, K.
B.; Miller, S. J. J. Am. Chem. Soc. 2008, 130, 16358. (4) Gustafson, J. L.; Sigman, M. S.; Miller, S. J. Org. Lett. 2010, 12, 2794. (5) Harper, K. C.; Sigman, M. S. Science 2011, 333, 1875. (6) Harper, K. C.; Sigman, M. S. Proceedings of the National Academy of Sciences 2011, 108, 2179. (7) Bott, G.; Field, L. D.; Sternhell, S. J. Am. Chem. Soc. 1980, 102, 5618. (8) Adams, R.; Yuan, H. C. Chem. Rev. 1933, 12, 261. (9) Winstein, S.; Holness, N. J. J. Am. Chem. Soc. 1955, 77, 5562. (10) Hansch, C.; Leo, A. Exploring QSAR: Fundamentals and Applications in Chemistry and Biology;
American Chemical Society: Washington, DC, 1995. (11) Charton, M. J. Am. Chem. Soc. 1975, 97, 3691. (12) Charton, M. J. Org. Chem. 1976, 41, 2217. (13) A. Verloop, J. T. In Bological Activity and Chemical Structure; Buisman, J. A., Ed.; Elsevier:
Amsterdam, 1977, p 63. (14) A. Verloop, J. T. In QSAR in Drug Desing and Toxicology; D. Hadzi, B. J.‐B., Ed.; Elsevier:
Amsterdam, 1987, p 97. (15) Verloop, A. In Drug Design; Ariens, E. J., Ed.; Academic Press: New York, 1976; Vol. III, p 133. (16) Verloop, A. In IUPAC Pesticide Chemistry; Miyamoto, J., Ed.; Pergamon: Oxford, 1983; Vol. 1, p
339.
© 2012 Macmillan Publishers Limited. All rights reserved.