Computational modelling of "PERspectives" has been proven to be crucial in order to accelerate and improve the wet lab tasks, aiming to achieve significant and reproducible results. Moreover, by setting the basis of the computational aspect of PERspectives, we further establish a potential workflow for future applications both on our as well as on relevant fields.
Our computational approach was based on modelling the senders and the recievers populations seprately. The aim of the model is to predict the change in concentrations of key chemical substances, the rate of gene expression and to evaluate the regulatory capacity of our system. Also, based on the experimental work carried out by the wet lab, we determined some kinetic constants to create a customised kinetic model for our system.
But we did not limit ourselves to just creating a kinetic model. For our system it is also important to observe the effect of lactone diffusion as well as the stability analysis.
In the following sections, we describe an outline of the aforementioned models and, most importantly, the thinking process we followed during their development. We hope they will serve as a helpful companion to anyone who wishes to use "PERspectives" in the future. As everything else in iGEM, our work is open to recommendations and improvements from the SynBio community.
Name | Translation Initiation Rate (au) | Strength |
---|---|---|
BCD2 | 287840.97 | Very Strong |
S4 | 85298.9 | Strong |
S5 | 78686.2 | Strong |
S10 | 31126.5 | Medium |
S11 | 30580.8 | Medium |
S12 | 11058.3 | Weak |
S13 | 10752.1 | Weak |
BCD1 | 9455.2 | Weak |
BCD8 | 2190.04 | Weak |
BSD12 | 12782.75 | Weak |
BCD14 | 22019.63 | Weak |
\(\frac{d[ATC_{int}]}{dt} = k_{-comlex}[complex]- k_{complex}[ATC_{int}][TetR] - d_{ATC}[ATC_{int}] \ (1) \)
\(\frac{d[tetR]}{dt} = \frac{p_{J23105} * k_{tetr} * CN_{senders}}{d_{mRNATetR}} + k_{-complex} [complex] - k_{complex} [ATC_{int}] [TetR] - d_{TetR} [TetR] \ (2) \)
\(\frac{d[complex]}{dt} = k_{complex}[ATC_{int}][TetR] - k_{-complex}[complex] - d_{complex}[complex] \ (3)\)
\(\frac{d[LuxI]}{dt} = \frac{p_{tet} * k_{LuxI} * CN_{senders}}{d_{mRNALuxI}} * [\frac{Kd_{tet}^n}{Kd_{tet}^n + [TetR]^n} - d_{LuxI} [LuxI]] \ (4) \)
\(\frac{d[AHL_{int}]}{dt} = k_{A} [LuxI] - d_{AHL} [AHL] \ (5)\)
Kinetic Constant | Description | Value | Source |
---|---|---|---|
CN | Plasmid Copy Number (Senders) | 10(Low) | Wet Lab Construct Used |
n | LuxI Hill Coefficient (in aTc-TetR system) | 3 | Team:HSiTAIWAN 2016 |
d_comlex | Degradation rate of aTcTetR | 0.025(1/min) | Literature |
d_mRNATetR | Degradation Rate of mRNA TetR | ln(2)/30 = 0.023(1/sec) | Literature |
dTetR | Degradation Rate of TetR | 0.631 [1/sec] | Literature |
d_AHL | Degradation Rate of AHL | 0.01 min-1 | Literature |
d_mRNALuxI | Degradation Rate of Luxi Mrna | 0.347 (1/min) | Literature |
d_LuxI | Degradation Rate of LuxI protein | 0.00167(1/min) | Team MIT 2010 |
d_aTc | Degradation Rate of aTc | ln(2)/20 (1hours) | Literature |
KA | Synthesis Rate of AHL by LuxI | 0.04(1/min) | Literature |
k-1_aTcTetR | Dissociation Rate of aTcTetR | 4.2*10^-4(1/min) | Literature |
k_aTcTetR | Synthesis Rate of aTcTetR | k-1/Kd | |
Kd_pTet | Dissociation constant of tetR with the pTet Promoter | 50 nM | Team:HSiTAIWAN 2016 |
Kd_aTcTetR | Dissociation constant of aTC with the tetR molecule | 15 nM | Team:HSiTAIWAN 2016 |
p_pTet | Transcription Rate of ptet promoter | 25.3*10^(-3) [1/sec] | Calculated using "Salis Lab" Calculator |
p_J23105 | Transcription Rate of J23105 promoter | 30*10^(-3) [1/sec] | Registry |
k_luxi | Translation Rate of LuxI | Taken from our RBS Library | Calculated using "Salis Lab" Calculator |
k_tetr | Translation Rate of TetR | 11*10^(-3)[1/sec] | Calculated using "Salis0.129*10 Lab" Calculator |
Nmin | Initial value of E.coli | 0.129*10^(7) | Calculated from lab data about OD |
Nmax | Maximum value of E.coli | 2.14*10^(7) | Calculated from lab data about OD |
Vbead | Volume of culture | 200 μL | From Wet Lab Setup |
\(\frac{dN}{dt} = \mu * [N] *(1-\frac {[N]}{[N_{max}]}) \ (1) \)
\(\frac{d[ATC_{ext}]}{dt} = D_{ATC} [N] ([ATC_{int}] - \frac{V_{E.Coli}}{V_{bead}} [ATC_{ext}]) - d_{ATC} [ATC_{ext}] \ (2) \)
\(\frac{d[ATC_{int}]}{dt} = D_{ATC} * (\frac{V_{E.Coli}}{V_{bead}} [ATC_{ext}] - [ATC_{int}]) + k_(-complex)[comlex] -k_{complex} [ATC_{int}][TetR] -d_{ATC} [ATC_{int}] \ (3) \)
\(\frac{d[tetR]}{dt} = \frac{p_{J23105} * k_{tetr} * CN_{senders}}{d_{mRNATetR}} + k_{-complex} [complex] - k_{complex} [ATC_{int}] [TetR] - d_{TetR} [TetR] \ (4) \)
\(\frac{d[complex]}{dt} = k_{complex}[ATC_{int}][TetR] - k_{-complex} [complex] - d_{complex} [complex] \ (5)\)
\(\frac{d[LuxI]}{dt} = \frac{p_{tet} * k_{LuxI} * CN_{senders}}{d_{mRNALuxI}} * (\frac{Kd_{tet}^n}{Kd_{tet}^n + [TetR]^n} - d_{LuxI} [LuxI]) \ (6) \)
\(\frac{d[AHL]}{dt} =k_{A} [LuxI] - d_{AHL} [AHL] \ (7) \)
\(\frac{d[AHL_{int}}{dt} = k_{A} [LuxI] + D_{AHL} ( \frac{V_{E.Coli}}{V_{bead}} [AHL_{ext}]) - d_{AHL} [AHL_{ext}] \ (8) \)
\(\frac{d[AHL_{ext}}{dt} = D_{AHL} [N] ( [AHL_{int}] \frac{V_{E.Coli}}{V_{bead}} [AHL_{ext}]) - d_{AHL} [AHL_{ext}] \ (9) \)
Symbol | Description | Literature Value | Source | Fitted Value |
---|---|---|---|---|
mi | Senders Dilution Rate | 0.019 [1/min] | Harvard Bionumbers | 0.032 [1/min] |
a_tetr | Production Rate of TetR | k_tetr*p_ptet=1.18[1/min] | Calculated with "SalisLab" Calculator | 10.22 |
b_ptet | Leakiness of Ptet promoter | 0 (%) | assumed | 20(%) |
a_luxi | Production Rate of LuxI (for BCD2) | k_luxi*p_luxi=312[1/min] | Calculated with "SalisLab" Calculator | 128 |
n | LuxI Hill Coefficient | 3 | Team:HSiTAIWAN 2016 | 2 |
kd_ptet | Dissociation Constant of TetR with the ptet promoter | 50nM | Team:HSiTAIWAN 2016 | 2.5nM |
The receivers population simulates the activation function of the perceptron algorithm. So it was really important for us to create constructs in a way that we would secure a steep activation of the production of the final fluorescent protein, mNeon Green. We had to design plasmids that would give the receivers cells the ability to express our protein very strongly only if the amount of lactone they receive from the senders exceeds a certain threshold. We decided to test two different constructs in order to check which one will present the best response.
\(\frac{d[AHL_{int}]}{dt} = k_{-1}*[Mon]+D_{AHL}*(\frac{V_{eColi}}{V_{bead}} *AHL_{ext} - AHL_{int}) -k_{1}*[AHL_{int}][LuxR] - d_{AHL} [AHL_{int}] \ (1)\)
\(\frac{d[LuxR]}{dt}= \frac{p_{J23105}*k_{LuxR}*CN_{rec}}{d_{mRNALuxR}} + k_{-1} [Mon] - k_{1}[LuxR][AHL_{int}] - d_{LuxR}[LuxR] \ (2)\)
\(\frac{Mon}{dt} = k{1}[LuxR][AHL_{int}] + 2k_{-2}[Dimer]-k_{-1}[Mon]-2k_{2}[Mon]^2 - d_{Mon}[Mon] \ (3)\)
\(\frac{Dimer}{dt}= k_{2}[Mon]^2 - k_{-2}[Dimer]-d_{Dimer}[Dimer] \ (4) \)
\(\frac{d_{mng}}{dt} = \frac{p_{Lux}*k_{mng}*CN_{rec}}{d_{mRNAmng}} * [\beta_{Lux}+(1-\beta_{Lux})*\frac{[Dimer]^n}{Kd_{Lux}^n+[Dimer]^n}] -d_{mng}[mng] \ (5) \)
\(\frac{dN}{dt} = \mu * [N] *(1-\frac {[N]}{[N_{max}]}) \ (1)\)
\(\frac{d_{AHL_{ext}}}{dt} = D_{AHL}*[N]*(AHL_{int}- \frac{V_{eColi}}{V_{bead}} *AHL_{ext}) - d_{AHL}[AHL_{ext}] \ (2)\)
\(\frac{d[AHL_{int}]}{dt} = k_{-1}*[Mon]+D_{AHL}*(\frac{V_{eColi}}{V_{bead}} *AHL_{ext} - AHL_{int}) -k_{1}*[AHL_{int}][LuxR] - d_{AHL} [AHL_{int}] \ (3) \)
\(\frac{d[LuxR]}{dt}= \frac{p_{J23105}*k_{LuxR}*CN_{rec}}{d_{mRNALuxR}} + k_{-1} [Mon] - k_{1}[LuxR][AHL_{int}] - d_{LuxR}[LuxR] \ (4)\)
\(\frac{d[Mon]}{dt}= k_{1} [LuxR][AHL_{int}]+2 k_{-2}[Dimer]-k_{-1}[Mon]^2 - d_{Mon}[Mon] \ (5)\)
\(\frac{d[Dimer]}{dt} = k_{2}[Mon]^2 - k_{-2} [Dimer] - d_{Dimer} [Dimer] \ (6) \)
\(\frac{d[mng]}{dt} = \frac{p_{Lux}*k_{mng}*CN_{rec}}{d_{mRNAmng}} + [\beta_{Lux} +(1-\beta_{Lux})* \frac{[Dimer]^n}{Kd_{Lux}^n+[Dimer]^n}] - d_{mng} [mng] \ (7)\)
\(\frac{d[AHL_{int}]}{dt} = k_{-1}[Mon] - k_{1} * [AHL_{int}][LuxR] - d_{AHL}[AHL_{int}] \ (1) \)
\(\frac{d[LuxR]}{dt} = \frac{p_{Lux}*k_{LuxR}*CN_{rec}}{d_{mRNALuxR}} * [(1-\beta_{Lux})+\beta_{Lux} *\frac{[Dimer]^n}{Kd_{Lux}^n+[Dimer]^n}] + k_{-1}[Mon]-k_{1}[LuxR][AHL_{int}] - d_{LuxR}[LuxR] \ (2) \)
\(\frac{d[Mon]}{dt} = k_{2} [Mon]^2 - k_{-2}[Dimer] - k_{-1}[Mon]-2k_{2}[Mon]^2 -d_{Mon}[Mon] \ (3) \)
\(\frac{d[Dimer]}{dt} = k_{2} [Mon]^2 - k_{-2}[Dimer] - d_{Dimer}[Dimer] \ (4)\)
\(\frac{d[phlf]}{dt} = \frac{p_{Lux}*k_{phlf}*CN_{rec}}{d_{mRNAphlf}} * [\beta_{Lux} + (1-\beta_{Lux}) * \frac{Kd_{Lux}^n}{Kd_{Lux}^n+[Dimer]^n} ]- d_{phlf}[phlf] \ (5)\)
\(Hill_{Lux}= \beta_{Lux} + (1-\beta_{Lux}) *\frac{[Dimer]^n}{Kd_{Lux}^n+[Dimer]^n} \)
\(Hill_{phlf}= \beta_{phlf} + (1-\beta_{Lux}) *\frac{Kd_{Lux}^n}{Kd_{Lux}^n+[Dimer]^n} \)
\(\frac{d[mng]}{dt} = \frac{p_{hybrid}*k_{mng}*CN_{rec}}{d_{mRNAmng}} *Hill_{Lux} * Hill_{phlf} - d_{mng} [mng] \ (6)\)
\(\frac{dN}{dt} = \mu * [N] *(1-\frac{[N]}{[N_{max}]}) \ (1)\)
\(\frac{d[AHL_{ext}]}{dt} = D_{AHL} *[N] *( AHL_{int} - \frac{V_{eColi}}{V_{bead}} *AHL_{ext}) - d_{AHL}[AHL_{ext}] \ (2)\)
\(\frac{d[AHL_{int}]}{dt} = k_{-1}*[Mon]+D_{AHL}*(\frac{V_{eColi}}{V_{bead}}*AHL_{ext} - AHL_{int}) -k_{1}*[AHL_{int}][LuxR] - d_{AHL} [AHL_{int}] \ (3) \)
\(\frac{d[LuxR]}{dt}= \frac{p_{Lux}*k_{LuxR}*CN_{rec}}{d_{mRNALuxR}} [\beta_{Lux} + (1-\beta_{Lux}) *\frac{[Dimer]^n}{Kd_{Lux}^n+[Dimer]^n}] + k_{-1} [Mon] - k_{1}[LuxR][AHL_{int}] - d_{LuxR}[LuxR] \ (4)\)
\(\frac{d[Mon]}{dt} = k_{1}[LuxR][AHL_{int}] + 2 k_{-2}[Dimer] - k_{-1} [Mon] -2k_{2}[Mon]^2 - d_{Mon} [Mon] \ (5)\)
\(\frac{d[Dimer]}{dt} = k_{2} [Mon]^2 - k_{-2} [Dimer] \ (6)\)
\(\frac{d[phlf]}{dt} = \frac{p_{Lux}*k_{phlf}*CN_{rec}}{d_{mRNAphlf}} * [\beta_{Lux} + (1-\beta_{Lux}) * \frac{Kd_{Lux}^n}{Kd_{Lux}^n+[Dimer]^n} ]- d_{phlf}[phlf] \ (7)\)
\(Hill_{Lux}= \beta_{Lux} + (1-\beta_{Lux}) *\frac{[Dimer]^n}{Kd_{Lux}^n+[Dimer]^n} \)
\(Hill_{phlf}= \beta_{phlf} + (1-\beta_{Lux}) *\frac{Kd_{Lux}^n}{Kd_{Lux}^n+[Dimer]^n} \)
\(\frac{d[mng]}{dt} = \frac{p_{hybrid}*k_{mng}*CN_{rec}}{d_{mRNAmng}} *Hill_{Lux} * Hill_{phlf} - d_{mng} [mng] \ (8)\)
Kinetic Constant | Description | Value | Source |
---|---|---|---|
CN | Plasmid Copy Number (Receivers) | 17(Medium) | Wet Lab Constructs |
n | Hill Coefficient | 1.5 | Estimated (before fit) |
k1 | Synthesis Rate of LuxRAHL | k-1/kd1 | |
k2 | Synthesis Rate of LuxRAHL2 | k-2/kd2 | |
k-1 | Dissociation Rate of Monomer (LuxR-AHL) | 10(1/min) | Literature |
k-2 | Dissociation Rate of of dimer (LuxR-AHL)2 | 1(1/min) | Literature |
kd2 | Dissociation Constant of Dimer (LuxR-AHL)2 | 20 Nm | Literature |
kd1 | Dissociation Constant of Monomer (LuxR-AHL) | 100 nM | Literature |
kdlux | Dissociation constant of (LuxR-AHL)2 to the plux promoter | 200nM | Literature |
d_RA2 | Degradation Rate of LuxRAHL2 | 0.017(1/min) | Team Valencia 2018 |
d_RA | Degradation Rate of LuxRAHL | 0.156(1/min) | Team Valencia 2018 |
d_Mng | Degradation Rate of Mng | 0.010(1/min) | Literature |
d_mRNAmNGt | Degradation Rate of Mrna_Mng | 0.039(1/min) | Literature |
d_LuxR | Degradation Rate of LuxR | 0.002(1/min) | Literature |
d_mRNALuxR | Degradation Rate of LuxR Mrna | 0.347(1/min) | Literature |
β_Lux | Basal Expression of pLux Promoter (as min/max) | 4.8%=(1043/21290) | Registry |
D_AHL | Diffusion Rate of External AHL through the membrane | 10(1/min) | Literature |
V_Ecoli | E.Coli Cell Volume | 1μ (m**3)=10**(-15) L | Literature |
μ | dilution rate | 0.019(1/min) | Literature |
Transc_Plux | Transcription Rate of LuxR promoter | 0.79/60(1/sec) | Team Valencia 2018 |
Vbead | Volume of culture | 200 μL | From Wet Lab Setup |
Nmin | Minimum value of cells in Bead | 0.29*10**(7) | From OD experimental Data |
Nmax | Maximum value of cells in Bead | 2.142*10^(7) | From OD experimental Data |
d_AHL | Degradation Rate of AHL | 0.01 min^-1 | Literature |
p_J23105 | Transcription Rate of constitutive promoter for LuxR | 58*10^(-3) (1/sec) | Registry |
p_Lux | Transcription Rate of LuxR Promoter for mneongreen | 0.79(1/min) | "Salis Lab" Calculator |
k_luxr | Translation Rate of LuxR (RBS BBa_B0034) | 18*10^(-3) (1/sec) | "Salis Lab" Calculator |
k_mneongreen | Translation Rate of mneongreen(RBS BBa_B0034) | 27*10^(-3) (1/sec) | "Salis Lab" Calculator |
Kinetic Constant | Description | Value | Source |
---|---|---|---|
p_lux | Transcription Rate of LuxR promoter for LuxR | 155*60*10^(-3) (1/sec) | "Salis Lab" Calculator |
k_luxr | Translation Rate of LuxR (RBS BBa_B0034) | 27*10^(-3) (1/sec) | "Salis Lab" Calculator |
p_lux_repr | Transcription Rate of LuxR_Repressible promoter for phlf | 3.69*10^(-3) (1/sec) | "Salis Lab" Calculator |
k_phlf | Translation Rate of Phlf (RBS BBa_B0034) | 4*10^(-3) (1/sec) | "Salis Lab" Calculator |
p_hybrid | Transcription Rate of Hybrid Promoter for mneongreen | 10*10^(-3) (1/sec) | "Salis Lab" Calculator |
k_mng | Translation Rate of mneongreen (RBS BBa_B0034) | 18.3*10^(-3) (1/sec) | "Salis Lab" Calculator |
d_mrnaphlf | Degradation Rate of mrna phlf | 0.02 (1/sec) | "Salis Lab" Calculator |
d_mrnaphlf | Degradation Rate of mrna phlf | 0.02 (1/sec) | Team SUSTech_Shenzhen 2021 |
d_phlf | Degradation Rate of phlf | 0.042(1/min) | Team Hong_Kong_HKUST 2017 |
Kd_phlf | Dissociation Constant for the phlf repressible promoter | 2*10^(-7) nM | Team Hong_Kong_HKUST 2017 |
Symbol | Description | Initial Value | Source | Fitted Value |
---|---|---|---|---|
n | Hill coefficient for OpLo | 1.5 | estimated | 0.5 |
b_plux | Leakiness of pLux promoter | 4.9% | Registry-BBa_R0062 | 10% |
a_luxr | Production Rate of LuxR | p_J23105*k_luxr=5.68 | SalisLab | 49.78 |
a_mng | Production Rate of mneongreen | p_plux*k_mng=0.85 | Salis Lab | 0.44 |
kd_lux | Dissociation Constant of LuxR Promoter | 200nM | Literature | 10nM |
d_mng | Degradation Rate of mneongreen | 0.01(1/min) | Literature | 0.023(1/min) |
d_mrna_mng | Degradation Rate of mrna mneongreen | 0.039(1/min) | Literature | 0.019(1/min) |
mi | Dilution Rate | 0.019(1/min) | Literature | 0.0028(1/min) |
Symbol | Description | Initial Value | Source | Fitted Value |
---|---|---|---|---|
a1 | Production Rate of LuxR | k_luxr*p_lux=15 | Salis Lab | 5.67 |
a2 | Production Rate of Phlf | k_phlf*p_lux_repr=0.053 | Salis Lab | 0.1 |
a3 | Production Rate of mneongreen | p_hybrid*k_mng=0.6588 | Salis Lab | 0.62 |
n | Hill Coefficient for PFR | 2 | Estimated | 2.31 |
b_phlf | Leakiness of phlf promoter | 10% | Estimated | 0.36 |
kd_phlf | Dissociation Constant of phlf with phlf promoter | 2*10^(-7)) | Team Hong_Kong_HKUST 2017 | 7.95*10^(-7) |
d_phlf | Degradation rate of phlf | 0.042(1/min) | Team Hong_Kong_HKUST 2017 | 0.03 |
d_mrna_phlf | Degradation rate of mrna_phld | 1.2(1/min) | Team SUSTech_Shenzhen 2021 | 0.06 |
\( a_{send} = a_{1} + a_{2} +a_{3}\)
\(\frac{d[x_{1}]}{dt} = f_{1}(t)= (a_{send}) * \frac{u^{n_{1}}}{K_{1}{^n_{1}}+u^{n_{1}}} - d_{1}* x_{1}\)
\(\frac{d[x_{2}]}{dt} = f_{2}(t)= (a_{rec}) * \frac{x_{1}{^n_{2}}}{K_{2}{^n_{2}} + x_{1}{^n_{2}}} - d_{2} * x_{2} \)
\( x_{dot}(t) = A * x(t) + B* u(t) \)
\(y_{dot}(t) = C * x(t) +D * u(t)\)
\(x(t) = [x_{1}(t),x_{2}(t)]\)
\(u(t)=aTc\)
\(x_{1}(t)= AHL\)
\(x_{2}(t) = out\)
\( where \ A=[a_{11}, a_{12}][a_{21}, a_{22}] \)
\(B=[b_{1}],b_{2}]\)
\(a_{11}= \frac{df_{1}}{dx_{1}} = -d_{1} \)
\(\frac{d[x_{2}]}{dt} = f_{2}(t)= (a_{send}) * \frac{u^n2}{K_{1}^n2+u^n2} - d_{2}* x_{2}\)
\(b_{1}=\frac{df_{1}}{du}=a_{send} = \frac{a_{send}*n_{1}*K_{1}^{n_{1}} * u(t_{0})^{n_{1}-1}}{(u(t_{0})^n_{1} + K_{1}^{n_{1}})^2}\)
\(a_{21}=\frac{df_{2}}{dx_{1}}=a_{rec} = \frac{a_{rec}*n_{2}*K_{2}^{n_{2}} * x_{1}(t_{0})^{n_{2}-1}}{(x_{1}(t_{0})^{n_{2}} + K_{2}^{n_{2}})^2} \)
\(a_{22} = \frac{df_{2}}{dx_{2}}=-d_{2}\)
\(b_{2} = \frac{df_{2}}{du}=0\)
\(λ_{1} = -d_{1} \ and \ λ_{2} = -d_{2}\)
\(H(s)= \frac{n1*n2*K_{1}^{n1}*K_{2}^{n2}*a_{rec}*a_{send}*x1(t0)^{n1}*u(t0)^{n1-1}}{x1(t0)*(s+d1)*(s+d2)} = \frac{A}{x1(t0)*(s+d1)*(s+d2)}\)
Symbol | Value |
---|---|
K1 | 50nM |
K2 | 200nM |
n1 | 2 |
n2 | 1.5 |
d1 | 0.01 [1/min] |
d2 | 0.01 [1/min] |
x1(t0) | 2*10^(-9) |
u(t0) | 10^(-6) |
a_rec | 0.85 |
a_send | 312 |
Thus combining RBSs with different strengths in populations that are induced or not, we create patterns that can be provided to the senders as inputs, which our system will be able to recognize. Theoretically, from each 3x1 bit pattern we expect a different amount of lactone production when all of the senders subpopulations are induced with aTc. Since the strength of each RBS is translated into an amount of lactone produced and finally the subpopulations of senders are mixed, the position of each bit does not matter but only what RBS each subpopulation has and if the population is induced or not. Furthermore, it is obvious that the maximum amount of the lactone will be produced in case no. 2 where the subpopulations of the senders all have very strong RBSs. Respectively, the lower amount of the lactone is expected in case no. 4 where the subpopulations of the senders all have very weak RBSs.
Our proposed implementation of the perceptron weights with the use of synthetic RBS sequences could become the template for the development of plug-and-play devices that enable the modification of the inputs' weights regarding the desired behaviour and application. We selected to use RBSs instead of promoters for the pattern recognition because of their versatility and it is quicker to and easier to predict the translation initiation rate of an RBS instead of directly evolutionize a promoter to have the desired strength. Overall, such a multicellular system has various applications in many scientific fields, from medicine to the environment and the thriving field of biocomputing. It could become the starting point for the development of “smart biosensors” or “smart drugs” and even perform complex computational tasks using much less energy than the amount needed by a computer, aiming to solve one of the most popular problems nowadays, the high cost of computational performances
After the characterization of our model, we wanted to simulate both the senders and the receivers populations and the way they would interact with each other. In this way, not only would we get a validity of our characterizations, but also make a direct comparison of the OpLo and PFR receivers. But at first we need to set up the pattern to be recognized as well as the RBS’s to be used.
For that purpose we decided to use only the sender’s constructs that were tested experimentally and accurately determine the a_luxi parameter of our model which depicts the strength of the AHL produced.
RBS | Value Wet | Value Dry | a_luxi_fitted |
---|---|---|---|
BCD2 | 43704 | 43680 | 128 |
BCD12 | 16981 | 16209 | 47.5 |
S11 | 34943 | 34125 | 100 |
S13 | 19903 | 20475 | 60 |
S4 | 5926 | 5972 | 17.5 |
S5 | 597 | 597 | 1.75 |
S10 | 17106 | 17062 | 50 |
S12 | 2887 | 2900 | 8.5 |
BCD8 | 4050 | 4095 | 12 |
BCD14 | 23202 | 23205 | 68 |
Also, the pattern simulated is presented below:
Index | Senders 1 | Senders 2 | Senders 3 | Desired Output |
---|---|---|---|---|
0 | 0 | 0 | 0 | 0 |
1 | 0 | 0 | 1 | 0 |
2 | 0 | 1 | 0 | 0 |
3 | 0 | 1 | 1 | 1 |
4 | 1 | 0 | 0 | 1 |
5 | 1 | 0 | 1 | 1 |
6 | 1 | 1 | 0 | 1 |
7 | 1 | 1 | 1 | 1 |
Via training from our software for the pattern above, it was concluded that the sequences to be used should be the following:
,meaning that we should use 1 strong RBS and two medium to low ones.
For the sender's 1 population we decided to use our strongest RBS, BCD2 due to the fact that we want output = 1 whenever the sender’s population is induced. As for the other 2 populations, we decided to use a medium RBS for each one, BCD12, since we would like to have no fluorescence when only each one of these sender’s population is induced (see index 1,2). On the other hand, when both senders 1 and 2 are induced we expect that the strength of two medium RBS combined would lead to an OC6 production similar to that of BCD2 and thus lead to high fluorescence.
It should be mentioned that we didn’t use synthetic 6 RBS-which was the result of the training on the application-for the simulation due to the lack of experimental data for that construct. However, the BCD12 RBS is also a relatively suitable candidate as a medium to low RBS, always in comparison to the strongest one, which is BCD2 (as shown in Table 1).
We performed a simulation of our system which can be found on the ‘proof_all.ipynb’ code on gitlab. In particular, we used a 1/7 senders/receivers cell culture ratio since we observed that it was the ideal ratio for our pattern recognition task. The culture volume remained 200μL. For more details on that you can take a look at our model analysis on the proof of concept page .
As we’ve seen on the characterization part of our model, the sender’s model takes as input the aTc concentration for induction and gives as output the molecules of external AHL. On the other hand, the receiver’s model uses concentrations of input OC6-instead of molecules-in order to estimate the output mneongreen. So, for that specific task, we added a simple expression to the implementation code that converts molecules of OC6 to Molarity:
Molarity_AHL=Molecules_AHL/(NA*V_bead), where:
NA =6.023*10**(23) is avogadro’s number
V_bead=200μL, the total cell volume
Other than that, we simulated the entire model system and received the following results:
From the results above, the significance of the steep activation function is displayed; for the high value patterns the AHL produced from BCD2 is enough to give us the wanted output 1. However, the analog nature of OpLo means that even the medium RBS alone can offer a fluorescence output similar to that of fully induced BCD2. On the other hand, such a concern doesn’t exist on the PFR construct, where the receivers are not activated only with the induction of one medium RBS. Moreover, we witness that in the pattern with index 2, 2 medium RBS’s are more than capable to induce the receiver’s population.