SlideShare a Scribd company logo
1 of 1
Download to read offline
COMPARTMENT-BASED MODEL FOR VI-B
PEPTIDE DEGRADATION
Nhat Pham1
, Brooks K. Emerick1
, Michelle L. Kovarik2
, Allison J. Tierney2
, Kunwei Yang2
1
Department of Mathematics,
2
Department of Chemistry, Trinity College
Abstract
A peptide is a chemical compound made up of amino acid bonds
which can be broken down by enzymes in the body, converting the
peptide into smaller chains. Knowing the kinetics of peptide degra-
dation can be useful in designing peptides that resist degradation. In
this research project, we are interested in finding the constant rates
of degradation of a peptide substrate reporter for protein kinase B
(VI-B) in five different cell cultures (Dictyostelium, HeLa, LNCaP,
Yeast, and E.Coli) using data collected by Prof. Kovarik, Allie Tier-
ney, and Kunwei Yang of the Chemistry Department. Based on
the physical process of the peptide, we see that compartment-based model can be used to track
the evolution of peptide strand as time evolves. We fit the solution functions of the model to the
given time-series data sequentially using tools of least-squares regression, and determine the best
fit parameters to be parameter values that yield the minimum residual errors among all trials.
Modeling results show variations in rate of degradation among reactions and organisms, reveal
which amino acid residues and fragments are targeted by peptidases and consequently the method
demonstrate potential utility in peptide substrate reporter design.
Model Formulation
VI-B are readily degraded by peptidases to break down into smaller fragments whose concentra-
tions are measured. In this work, we use compartment-based model to describe the concentration
over time. The model includes several assumptions:
(1) the initial reporter is the only source of fragments
(2) the parent peptide or any small fragments do not leave the system, but are simply converted
to smaller fragments
(3) any larger peptide could be cleaved to form any smaller peptide, but only fragments re-
taining 6-FAM label will be detected
(4) the system follows first-order kinetics and all rate constants are non-negative
Fig.1 Model of reporter (B) metabolism into shorter fragments
from eukaryotic cells with rate constants ki
Fig.2 Reporter concentration as a function of time
The model can be translated into a series of differential equations:
dB
dt
= −(k8 + k7 + k6 + k5)B = −dpB (1)
dB8
dt
= −(k87 + k86 + k85)B8 + k8R = −d8B8 + k8B (2)
dB7
dt
= −(k76 + k75)B7 + k87B8 + k7B = −d7B7 + k87B8 + k7B (3)
dB6
dt
= −k65B6 + k76B7 + k86B8 + k6B (4)
dB5
dt
= k65B6 + k75B7 + k85B8 + k5B, (5)
In E.Coli lysates, only segments VI-B7 through VI-B4 are observed. As a result, a slightly different
compartment-based model was required to describe this system. This model is mathematically
equivalent to the model above, and the same assumptions were applied.
Data Fitting Methods
To find the best fit parameters for each segment, we first solved Equations (1)-(5) using basic
techniques for solving differential equations, then match the explicit solutions to the given time-
series data in the least-squares sense. We first found the best fit parameter for the parent peptide,
B(t), then found best fit parameters for B8, B7, B6, B5 sequentially .
• Fitting data for the reporter (B):
- Find explicit solution to equation (1):
B(t, dp) = e−dpt
- Linearize the function by taking the natural
logarithm ln(B) = −dpt and match it to the
logarithm of the given data.
- Minimize the residual error using the least-
squares regression line and find the best fit for
parameter dp
0 10 20 30 40 50 60
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
time (min)
ConcentrationProportion
Data-Fitting (VI-B,LNCaP)
• Fitting data for B8:
- Substitute best fit value for dp to Eq. (2) and solve for B8:
B8(t) = k8
dp−k8
(e−dpt − e−dpt)
- Fit function to the data using non-linear least-squares approach:
+ Set lowerbound and upperbound for parameters: 0 ≤ d8 ≤ 1 and 0 ≤ k8 ≤ dp
+ Run 10,000 simulations in Matlab lsqnonlin function with randomized initial conditions be-
tween [0,1] and report the parameter value that yields the minimum residual error to be best
fit parameter, also considering the most popular value among trials.
0.0657 0.0657 0.0658 0.0658 0.0659 0.0659 0.066
0
1000
2000
3000
4000
Distribution for Parameter d8 (LNCaP)
d8 fit
Frequency
10000 Trials
Min Res = 3.2052e-05
Best d8 = 0.065783
−1.1824 −1.1822 −1.182 −1.1818 −1.1816 −1.1814 −1.1812 −1.181 −1.1808 −1.1806
0
1000
2000
3000
4000
Log Distribution for Parameter d8 (LNCaP)
log(d8) fit
Frequency
10000 Trials
Min Res = 3.2052e-05
Best log(d8) = -1.1819
5.288 5.29 5.292 5.294 5.296 5.298 5.3
x 10
−3
0
1000
2000
3000
4000
Distribution for Parameter k8 (LNCaP)
k8 fit
Frequency
10000 Trials
Min Res = 3.2052e-05
Best k8 = 0.005291
−2.2767 −2.2766 −2.2765 −2.2764 −2.2763 −2.2762 −2.2761 −2.276 −2.2759 −2.2758 −2.2757
0
1000
2000
3000
4000
Log Distribution for Parameter k8 (LNCaP)
log(k8) fit
Frequency
10000 Trials
Min Res = 3.2052e-05
Best log(k8) = -2.2765
• Fitting data for other fragments: Applying similar method to find other best-fit pa-
rameters. Note that B5(t) is completely determined by the previous four solutions.
0 50 100 150 200 250 300 350
0
0.1
0.2
0.3
0.4
0.5
0.6
time (min)
ConcentrationProportion
Data-Fitting (Dictyostelium)
VI−B5
VI−B6
VI−B7
VI−B8
0 10 20 30 40 50 60
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
time (min)
ConcentrationProportion
Data-Fitting (LNCaP)
VI−B5
VI−B6
VI−B7
VI−B8
0 10 20 30 40 50 60
−0.05
0
0.05
0.1
0.15
0.2
0.25
0.3
time (min)
ConcentrationProportion
Data-Fitting (HeLa)
VI−B5
VI−B6
VI−B7
VI−B8
0 50 100 150 200 250 300 350
−0.05
0
0.05
0.1
0.15
0.2
0.25
time (min)
ConcentrationProportion
Data-Fitting (Yeast)
VI−B5
VI−B6
VI−B7
VI−B8
Best-fit Parameters and Errors
Table 1: Best fit parameter values with minimum residual and highest frequency
Parameter
Dictyostelium HeLa LNCaP Yeast E-Coli
Min Res Max Freq Min Res Max Freq Min Res Max Freq Min Res Max Freq Min Res** Max Freq
dp 7.3783e-03 - 7.3096e-03 - 4.4455e-02 - 2.3341e-03 - 2.1500e-02 -
d8 4.2125e-03 4.36e-03 4.4931e-02 4.49e-02 6.5783e-02 6.58e-02 1.7033e-02 1.71e-02 4.3809e-04 4.38e-04
k8 2.2987e-04 2.33e-04 2.8461e-03 2.85e-03 5.2909e-03 5.29e-03 1.5109e-03 1.51e-03 1.2944e-02 1.29e-02
d7 3.9124e-04 3.91e-04 3.0218e-02 3.46e-02 1.0075e-01 1.11e-01 9.4708e-03 9.47e-03 1.8829e-02 1.88e-02
k87 1.2876e-13 2.22e-14 2.0177e-02 2.28e-02 5.4015e-02 4.40e-02 1.7033e-02 1.70e-02 3.3537e-06 3.37e-06
k7 5.4568e-03 5.46e-03 2.3930e-04 2.06e-04 4.7989e-04 6.48e-04 1.7692e-04 1.78e-04 8.5558e-03 8.56e-03
d6 3.2309e-12 6.48e-07 1.4341e-02 1.61e-02 1.0000e-00 1.00e-00 1.5530e-09 1.73e-07 4.6765e-13 9.00e-09
k86 4.2125e-03 4.21e-03 4.3660e-08 2.24e-14 1.3304e-11 2.22e-14 9.7969e-15 9.80e-15 4.3474e-04 4.35e-04
k76 1.2180e-11 2.25e-14 2.8988e-02 3.01e-02 2.3307e-14 2.60e-05 5.0344e-03 5.03e-03 2.2205e-14 2.22e-14
k6 1.5388e-03 1.54e-03 1.9253e-04 2.56e-04 1.8884e-02 1.89e-02 2.2303e-14 2.22e-14 8.4237e-08 8.42e-08
k85 2.2204e-14 - 2.4754e-02 - 1.1768e-02 - 2.2232e-14 - 2.2205e-14 -
k75 3.9124e-04 - 1.2300e-03 - 1.0075e-01 - 4.4364e-03 - 1.8829e-02 -
k65 3.2309e-12 - 1.4341e-02 - 1.0000e-00 - 1.5530e-09 - 4.6765e-13 -
k5 1.5282e-04 - 4.0318e-03 - 1.9801e-02 - 6.4625e-04 - 2.2206e-14 -
**EColi’s parameter sequence:dp,d7, k7, d6,k76,k6,d5,k75,k65,k5,k74,k64,k54
* Min Res and Max Feq values with more than two orders of magnitude difference are in red
Table 2: Minimum residuals for each peptide and species
Segment Dictyostelium HeLa LNCaP Yeast E-Coli
VI-B 9.3214e-02 2.1014e-02 7.9433e-02 1.1629e-01 1.4732e-01
VI-B8 2.7728e-04 9.7549e-05 3.2052e-05 3.0242e-04 -
VI-B7 2.3038e-03 6.5266e-06 1.1345e-05 6.8960e-04 2.3609e-02
VI-B6 4.2415e-03 2.2384e-04 4.6258e-03 1.6917e-02 1.2882e-03
VI-B5 4.2415e-03 2.2384e-04 4.6258e-03 1.6917e-02 1.9770e-01
VI-B4 - - - - 1.9770e-01
General Model
The simplified general model for the system of ODEs written in matrix form:
dy
dt
=






−α1 0 0 0 0
α1/4 −α2 0 0 0
α1/4 α2/3 −α3 0 0
α1/4 α2/3 α3/2 −α4 0
α1/4 α2/3 α3/2 α4 0






y, y(0) =






1
0
0
0
0






.
General solution: y(t) = C1v1e−α1t + C2v2e−α2t + C3v3e−α3t + C4v4e−α4t + C5v5 , where
v1 =








4(α1−α2)
4α2−3α1
3(α1−α3)
3α3−2α1
2(α1−α4)
2α4−α1
α1
4α2−3α1
−3(α1−α3)
3α3−2α1
2(α1−α4)
2α4−α1
α1
3α3−2α1
2(α1−α4)
2α4−α1
−α1
2α4−α1
1








, v2 =







0
−3(α2−α3)
3α3−2α2
2(α2−α4)
2α4−α2
α2
3α3−2α2
2(α2−α4)
2α4−α2
−α2
2α4−α2
1







, v3 =






0
0
2(α3−α4)
2α4−α3
−α3
2α4−α3
1






, v4 =




0
0
0
−1
1



 , v5 =




0
0
0
0
1




• Future work: We would like to generalize the compartment-based model to any number of
peptide fragments - to develop the general solution to this ODE system in n-dimension.
Conclusions
• Rate constants for the degradation reactions varied widely between reactions and organisms.
Overall, the degradation resistance of VI-B suggests that it is sufficiently stable for application
across eukaryotic cells (with the exception of the rapid degradation in LNCaP lysates).
• Given only ten data points for each peptide strand, we found it more economical to fit the data
in a sequential order rather than all at the same time and the sequential method described
above yielded smaller residual norms.
• These results demonstrate the potential utility of compartment-based models for optimizing
substrate reporters for degradation resistance.
[1] Wu D. , Sylvester J. E. , Parker L. L. , Zhou G. , Kron S. J. 2010 Peptide reporters of kinase activity in whole cell
lysates. Biopolymers 94:475486.
[2] Phillips R. M. , Bair E. , Lawrence D. S. , Sims C. E. , Allbritton N. L. 2013 Measurement of protein tyrosine
phosphatase activity in single cells by capillary electrophoresis. Anal Chem 85:61366142.
[3] Ng E. X. , Miller M. A. , Jing T. , Chen C. H. 2016 Single cell multiplexed assay for proteolytic activity using droplet
microfluidics. Biosensors and Bioelectronics 81:408414.

More Related Content

Similar to Pham,Nhat_ResearchPoster

OpenCL applications in genomics
OpenCL applications in genomicsOpenCL applications in genomics
OpenCL applications in genomicsUSC
 
Utlization Cat Swarm Optimization Algorithm for Selected Harmonic Elemination...
Utlization Cat Swarm Optimization Algorithm for Selected Harmonic Elemination...Utlization Cat Swarm Optimization Algorithm for Selected Harmonic Elemination...
Utlization Cat Swarm Optimization Algorithm for Selected Harmonic Elemination...IJPEDS-IAES
 
System overflow blocking-transients-for-queues-with-batch-arrivals-using-a-fa...
System overflow blocking-transients-for-queues-with-batch-arrivals-using-a-fa...System overflow blocking-transients-for-queues-with-batch-arrivals-using-a-fa...
System overflow blocking-transients-for-queues-with-batch-arrivals-using-a-fa...Cemal Ardil
 
Universal approximators for Direct Policy Search in multi-purpose water reser...
Universal approximators for Direct Policy Search in multi-purpose water reser...Universal approximators for Direct Policy Search in multi-purpose water reser...
Universal approximators for Direct Policy Search in multi-purpose water reser...Andrea Castelletti
 
Application of parallel hierarchical matrices and low-rank tensors in spatial...
Application of parallel hierarchical matrices and low-rank tensors in spatial...Application of parallel hierarchical matrices and low-rank tensors in spatial...
Application of parallel hierarchical matrices and low-rank tensors in spatial...Alexander Litvinenko
 
Computational tools for drug discovery
Computational tools for drug discoveryComputational tools for drug discovery
Computational tools for drug discoveryEszter Szabó
 
ECONOMIC LOAD DISPATCH USING PARTICLE SWARM OPTIMIZATION
ECONOMIC LOAD DISPATCH USING PARTICLE SWARM OPTIMIZATIONECONOMIC LOAD DISPATCH USING PARTICLE SWARM OPTIMIZATION
ECONOMIC LOAD DISPATCH USING PARTICLE SWARM OPTIMIZATIONMln Phaneendra
 
ISFragkopoulos - Seminar on Electrochemical Promotion
ISFragkopoulos - Seminar on Electrochemical PromotionISFragkopoulos - Seminar on Electrochemical Promotion
ISFragkopoulos - Seminar on Electrochemical PromotionIoannis S. Fragkopoulos
 
Metaheuristic Tuning of Type-II Fuzzy Inference System for Data Mining
Metaheuristic Tuning of Type-II Fuzzy Inference System for Data MiningMetaheuristic Tuning of Type-II Fuzzy Inference System for Data Mining
Metaheuristic Tuning of Type-II Fuzzy Inference System for Data MiningVarun Ojha
 
Myers_SIAMCSE15
Myers_SIAMCSE15Myers_SIAMCSE15
Myers_SIAMCSE15Karen Pao
 
hERG SOT Poster 2010
hERG SOT Poster 2010hERG SOT Poster 2010
hERG SOT Poster 2010ShiminWang
 
hERG SOT Poster 2010
hERG SOT Poster 2010hERG SOT Poster 2010
hERG SOT Poster 2010karenbernards
 
Q pcr symposium2007-pcrarray
Q pcr symposium2007-pcrarrayQ pcr symposium2007-pcrarray
Q pcr symposium2007-pcrarrayElsa von Licy
 
LSSC2011 Optimization of intermolecular interaction potential energy paramete...
LSSC2011 Optimization of intermolecular interaction potential energy paramete...LSSC2011 Optimization of intermolecular interaction potential energy paramete...
LSSC2011 Optimization of intermolecular interaction potential energy paramete...Dragan Sahpaski
 
InstructionsACTCOEFF.XLSThis workbook will calculate activities, a.docx
InstructionsACTCOEFF.XLSThis workbook will calculate activities, a.docxInstructionsACTCOEFF.XLSThis workbook will calculate activities, a.docx
InstructionsACTCOEFF.XLSThis workbook will calculate activities, a.docxdirkrplav
 
InstructionsACTCOEFF.XLSThis workbook will calculate activities, a.docx
InstructionsACTCOEFF.XLSThis workbook will calculate activities, a.docxInstructionsACTCOEFF.XLSThis workbook will calculate activities, a.docx
InstructionsACTCOEFF.XLSThis workbook will calculate activities, a.docxmaoanderton
 

Similar to Pham,Nhat_ResearchPoster (20)

OpenCL applications in genomics
OpenCL applications in genomicsOpenCL applications in genomics
OpenCL applications in genomics
 
Utlization Cat Swarm Optimization Algorithm for Selected Harmonic Elemination...
Utlization Cat Swarm Optimization Algorithm for Selected Harmonic Elemination...Utlization Cat Swarm Optimization Algorithm for Selected Harmonic Elemination...
Utlization Cat Swarm Optimization Algorithm for Selected Harmonic Elemination...
 
System overflow blocking-transients-for-queues-with-batch-arrivals-using-a-fa...
System overflow blocking-transients-for-queues-with-batch-arrivals-using-a-fa...System overflow blocking-transients-for-queues-with-batch-arrivals-using-a-fa...
System overflow blocking-transients-for-queues-with-batch-arrivals-using-a-fa...
 
Universal approximators for Direct Policy Search in multi-purpose water reser...
Universal approximators for Direct Policy Search in multi-purpose water reser...Universal approximators for Direct Policy Search in multi-purpose water reser...
Universal approximators for Direct Policy Search in multi-purpose water reser...
 
M4L12.ppt
M4L12.pptM4L12.ppt
M4L12.ppt
 
Application of parallel hierarchical matrices and low-rank tensors in spatial...
Application of parallel hierarchical matrices and low-rank tensors in spatial...Application of parallel hierarchical matrices and low-rank tensors in spatial...
Application of parallel hierarchical matrices and low-rank tensors in spatial...
 
Computational tools for drug discovery
Computational tools for drug discoveryComputational tools for drug discovery
Computational tools for drug discovery
 
ECONOMIC LOAD DISPATCH USING PARTICLE SWARM OPTIMIZATION
ECONOMIC LOAD DISPATCH USING PARTICLE SWARM OPTIMIZATIONECONOMIC LOAD DISPATCH USING PARTICLE SWARM OPTIMIZATION
ECONOMIC LOAD DISPATCH USING PARTICLE SWARM OPTIMIZATION
 
ASHI2013HLA(1)
ASHI2013HLA(1)ASHI2013HLA(1)
ASHI2013HLA(1)
 
ACS 22LIE12 lab Manul.docx
ACS 22LIE12 lab Manul.docxACS 22LIE12 lab Manul.docx
ACS 22LIE12 lab Manul.docx
 
ISFragkopoulos - Seminar on Electrochemical Promotion
ISFragkopoulos - Seminar on Electrochemical PromotionISFragkopoulos - Seminar on Electrochemical Promotion
ISFragkopoulos - Seminar on Electrochemical Promotion
 
Rare B strangeness decay
Rare B strangeness decayRare B strangeness decay
Rare B strangeness decay
 
Metaheuristic Tuning of Type-II Fuzzy Inference System for Data Mining
Metaheuristic Tuning of Type-II Fuzzy Inference System for Data MiningMetaheuristic Tuning of Type-II Fuzzy Inference System for Data Mining
Metaheuristic Tuning of Type-II Fuzzy Inference System for Data Mining
 
Myers_SIAMCSE15
Myers_SIAMCSE15Myers_SIAMCSE15
Myers_SIAMCSE15
 
hERG SOT Poster 2010
hERG SOT Poster 2010hERG SOT Poster 2010
hERG SOT Poster 2010
 
hERG SOT Poster 2010
hERG SOT Poster 2010hERG SOT Poster 2010
hERG SOT Poster 2010
 
Q pcr symposium2007-pcrarray
Q pcr symposium2007-pcrarrayQ pcr symposium2007-pcrarray
Q pcr symposium2007-pcrarray
 
LSSC2011 Optimization of intermolecular interaction potential energy paramete...
LSSC2011 Optimization of intermolecular interaction potential energy paramete...LSSC2011 Optimization of intermolecular interaction potential energy paramete...
LSSC2011 Optimization of intermolecular interaction potential energy paramete...
 
InstructionsACTCOEFF.XLSThis workbook will calculate activities, a.docx
InstructionsACTCOEFF.XLSThis workbook will calculate activities, a.docxInstructionsACTCOEFF.XLSThis workbook will calculate activities, a.docx
InstructionsACTCOEFF.XLSThis workbook will calculate activities, a.docx
 
InstructionsACTCOEFF.XLSThis workbook will calculate activities, a.docx
InstructionsACTCOEFF.XLSThis workbook will calculate activities, a.docxInstructionsACTCOEFF.XLSThis workbook will calculate activities, a.docx
InstructionsACTCOEFF.XLSThis workbook will calculate activities, a.docx
 

Pham,Nhat_ResearchPoster

  • 1. COMPARTMENT-BASED MODEL FOR VI-B PEPTIDE DEGRADATION Nhat Pham1 , Brooks K. Emerick1 , Michelle L. Kovarik2 , Allison J. Tierney2 , Kunwei Yang2 1 Department of Mathematics, 2 Department of Chemistry, Trinity College Abstract A peptide is a chemical compound made up of amino acid bonds which can be broken down by enzymes in the body, converting the peptide into smaller chains. Knowing the kinetics of peptide degra- dation can be useful in designing peptides that resist degradation. In this research project, we are interested in finding the constant rates of degradation of a peptide substrate reporter for protein kinase B (VI-B) in five different cell cultures (Dictyostelium, HeLa, LNCaP, Yeast, and E.Coli) using data collected by Prof. Kovarik, Allie Tier- ney, and Kunwei Yang of the Chemistry Department. Based on the physical process of the peptide, we see that compartment-based model can be used to track the evolution of peptide strand as time evolves. We fit the solution functions of the model to the given time-series data sequentially using tools of least-squares regression, and determine the best fit parameters to be parameter values that yield the minimum residual errors among all trials. Modeling results show variations in rate of degradation among reactions and organisms, reveal which amino acid residues and fragments are targeted by peptidases and consequently the method demonstrate potential utility in peptide substrate reporter design. Model Formulation VI-B are readily degraded by peptidases to break down into smaller fragments whose concentra- tions are measured. In this work, we use compartment-based model to describe the concentration over time. The model includes several assumptions: (1) the initial reporter is the only source of fragments (2) the parent peptide or any small fragments do not leave the system, but are simply converted to smaller fragments (3) any larger peptide could be cleaved to form any smaller peptide, but only fragments re- taining 6-FAM label will be detected (4) the system follows first-order kinetics and all rate constants are non-negative Fig.1 Model of reporter (B) metabolism into shorter fragments from eukaryotic cells with rate constants ki Fig.2 Reporter concentration as a function of time The model can be translated into a series of differential equations: dB dt = −(k8 + k7 + k6 + k5)B = −dpB (1) dB8 dt = −(k87 + k86 + k85)B8 + k8R = −d8B8 + k8B (2) dB7 dt = −(k76 + k75)B7 + k87B8 + k7B = −d7B7 + k87B8 + k7B (3) dB6 dt = −k65B6 + k76B7 + k86B8 + k6B (4) dB5 dt = k65B6 + k75B7 + k85B8 + k5B, (5) In E.Coli lysates, only segments VI-B7 through VI-B4 are observed. As a result, a slightly different compartment-based model was required to describe this system. This model is mathematically equivalent to the model above, and the same assumptions were applied. Data Fitting Methods To find the best fit parameters for each segment, we first solved Equations (1)-(5) using basic techniques for solving differential equations, then match the explicit solutions to the given time- series data in the least-squares sense. We first found the best fit parameter for the parent peptide, B(t), then found best fit parameters for B8, B7, B6, B5 sequentially . • Fitting data for the reporter (B): - Find explicit solution to equation (1): B(t, dp) = e−dpt - Linearize the function by taking the natural logarithm ln(B) = −dpt and match it to the logarithm of the given data. - Minimize the residual error using the least- squares regression line and find the best fit for parameter dp 0 10 20 30 40 50 60 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 time (min) ConcentrationProportion Data-Fitting (VI-B,LNCaP) • Fitting data for B8: - Substitute best fit value for dp to Eq. (2) and solve for B8: B8(t) = k8 dp−k8 (e−dpt − e−dpt) - Fit function to the data using non-linear least-squares approach: + Set lowerbound and upperbound for parameters: 0 ≤ d8 ≤ 1 and 0 ≤ k8 ≤ dp + Run 10,000 simulations in Matlab lsqnonlin function with randomized initial conditions be- tween [0,1] and report the parameter value that yields the minimum residual error to be best fit parameter, also considering the most popular value among trials. 0.0657 0.0657 0.0658 0.0658 0.0659 0.0659 0.066 0 1000 2000 3000 4000 Distribution for Parameter d8 (LNCaP) d8 fit Frequency 10000 Trials Min Res = 3.2052e-05 Best d8 = 0.065783 −1.1824 −1.1822 −1.182 −1.1818 −1.1816 −1.1814 −1.1812 −1.181 −1.1808 −1.1806 0 1000 2000 3000 4000 Log Distribution for Parameter d8 (LNCaP) log(d8) fit Frequency 10000 Trials Min Res = 3.2052e-05 Best log(d8) = -1.1819 5.288 5.29 5.292 5.294 5.296 5.298 5.3 x 10 −3 0 1000 2000 3000 4000 Distribution for Parameter k8 (LNCaP) k8 fit Frequency 10000 Trials Min Res = 3.2052e-05 Best k8 = 0.005291 −2.2767 −2.2766 −2.2765 −2.2764 −2.2763 −2.2762 −2.2761 −2.276 −2.2759 −2.2758 −2.2757 0 1000 2000 3000 4000 Log Distribution for Parameter k8 (LNCaP) log(k8) fit Frequency 10000 Trials Min Res = 3.2052e-05 Best log(k8) = -2.2765 • Fitting data for other fragments: Applying similar method to find other best-fit pa- rameters. Note that B5(t) is completely determined by the previous four solutions. 0 50 100 150 200 250 300 350 0 0.1 0.2 0.3 0.4 0.5 0.6 time (min) ConcentrationProportion Data-Fitting (Dictyostelium) VI−B5 VI−B6 VI−B7 VI−B8 0 10 20 30 40 50 60 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 time (min) ConcentrationProportion Data-Fitting (LNCaP) VI−B5 VI−B6 VI−B7 VI−B8 0 10 20 30 40 50 60 −0.05 0 0.05 0.1 0.15 0.2 0.25 0.3 time (min) ConcentrationProportion Data-Fitting (HeLa) VI−B5 VI−B6 VI−B7 VI−B8 0 50 100 150 200 250 300 350 −0.05 0 0.05 0.1 0.15 0.2 0.25 time (min) ConcentrationProportion Data-Fitting (Yeast) VI−B5 VI−B6 VI−B7 VI−B8 Best-fit Parameters and Errors Table 1: Best fit parameter values with minimum residual and highest frequency Parameter Dictyostelium HeLa LNCaP Yeast E-Coli Min Res Max Freq Min Res Max Freq Min Res Max Freq Min Res Max Freq Min Res** Max Freq dp 7.3783e-03 - 7.3096e-03 - 4.4455e-02 - 2.3341e-03 - 2.1500e-02 - d8 4.2125e-03 4.36e-03 4.4931e-02 4.49e-02 6.5783e-02 6.58e-02 1.7033e-02 1.71e-02 4.3809e-04 4.38e-04 k8 2.2987e-04 2.33e-04 2.8461e-03 2.85e-03 5.2909e-03 5.29e-03 1.5109e-03 1.51e-03 1.2944e-02 1.29e-02 d7 3.9124e-04 3.91e-04 3.0218e-02 3.46e-02 1.0075e-01 1.11e-01 9.4708e-03 9.47e-03 1.8829e-02 1.88e-02 k87 1.2876e-13 2.22e-14 2.0177e-02 2.28e-02 5.4015e-02 4.40e-02 1.7033e-02 1.70e-02 3.3537e-06 3.37e-06 k7 5.4568e-03 5.46e-03 2.3930e-04 2.06e-04 4.7989e-04 6.48e-04 1.7692e-04 1.78e-04 8.5558e-03 8.56e-03 d6 3.2309e-12 6.48e-07 1.4341e-02 1.61e-02 1.0000e-00 1.00e-00 1.5530e-09 1.73e-07 4.6765e-13 9.00e-09 k86 4.2125e-03 4.21e-03 4.3660e-08 2.24e-14 1.3304e-11 2.22e-14 9.7969e-15 9.80e-15 4.3474e-04 4.35e-04 k76 1.2180e-11 2.25e-14 2.8988e-02 3.01e-02 2.3307e-14 2.60e-05 5.0344e-03 5.03e-03 2.2205e-14 2.22e-14 k6 1.5388e-03 1.54e-03 1.9253e-04 2.56e-04 1.8884e-02 1.89e-02 2.2303e-14 2.22e-14 8.4237e-08 8.42e-08 k85 2.2204e-14 - 2.4754e-02 - 1.1768e-02 - 2.2232e-14 - 2.2205e-14 - k75 3.9124e-04 - 1.2300e-03 - 1.0075e-01 - 4.4364e-03 - 1.8829e-02 - k65 3.2309e-12 - 1.4341e-02 - 1.0000e-00 - 1.5530e-09 - 4.6765e-13 - k5 1.5282e-04 - 4.0318e-03 - 1.9801e-02 - 6.4625e-04 - 2.2206e-14 - **EColi’s parameter sequence:dp,d7, k7, d6,k76,k6,d5,k75,k65,k5,k74,k64,k54 * Min Res and Max Feq values with more than two orders of magnitude difference are in red Table 2: Minimum residuals for each peptide and species Segment Dictyostelium HeLa LNCaP Yeast E-Coli VI-B 9.3214e-02 2.1014e-02 7.9433e-02 1.1629e-01 1.4732e-01 VI-B8 2.7728e-04 9.7549e-05 3.2052e-05 3.0242e-04 - VI-B7 2.3038e-03 6.5266e-06 1.1345e-05 6.8960e-04 2.3609e-02 VI-B6 4.2415e-03 2.2384e-04 4.6258e-03 1.6917e-02 1.2882e-03 VI-B5 4.2415e-03 2.2384e-04 4.6258e-03 1.6917e-02 1.9770e-01 VI-B4 - - - - 1.9770e-01 General Model The simplified general model for the system of ODEs written in matrix form: dy dt =       −α1 0 0 0 0 α1/4 −α2 0 0 0 α1/4 α2/3 −α3 0 0 α1/4 α2/3 α3/2 −α4 0 α1/4 α2/3 α3/2 α4 0       y, y(0) =       1 0 0 0 0       . General solution: y(t) = C1v1e−α1t + C2v2e−α2t + C3v3e−α3t + C4v4e−α4t + C5v5 , where v1 =         4(α1−α2) 4α2−3α1 3(α1−α3) 3α3−2α1 2(α1−α4) 2α4−α1 α1 4α2−3α1 −3(α1−α3) 3α3−2α1 2(α1−α4) 2α4−α1 α1 3α3−2α1 2(α1−α4) 2α4−α1 −α1 2α4−α1 1         , v2 =        0 −3(α2−α3) 3α3−2α2 2(α2−α4) 2α4−α2 α2 3α3−2α2 2(α2−α4) 2α4−α2 −α2 2α4−α2 1        , v3 =       0 0 2(α3−α4) 2α4−α3 −α3 2α4−α3 1       , v4 =     0 0 0 −1 1     , v5 =     0 0 0 0 1     • Future work: We would like to generalize the compartment-based model to any number of peptide fragments - to develop the general solution to this ODE system in n-dimension. Conclusions • Rate constants for the degradation reactions varied widely between reactions and organisms. Overall, the degradation resistance of VI-B suggests that it is sufficiently stable for application across eukaryotic cells (with the exception of the rapid degradation in LNCaP lysates). • Given only ten data points for each peptide strand, we found it more economical to fit the data in a sequential order rather than all at the same time and the sequential method described above yielded smaller residual norms. • These results demonstrate the potential utility of compartment-based models for optimizing substrate reporters for degradation resistance. [1] Wu D. , Sylvester J. E. , Parker L. L. , Zhou G. , Kron S. J. 2010 Peptide reporters of kinase activity in whole cell lysates. Biopolymers 94:475486. [2] Phillips R. M. , Bair E. , Lawrence D. S. , Sims C. E. , Allbritton N. L. 2013 Measurement of protein tyrosine phosphatase activity in single cells by capillary electrophoresis. Anal Chem 85:61366142. [3] Ng E. X. , Miller M. A. , Jing T. , Chen C. H. 2016 Single cell multiplexed assay for proteolytic activity using droplet microfluidics. Biosensors and Bioelectronics 81:408414.