Model

Bioreactor Modelling. . The scope of this project necessitated modeling parameters like cell, product, and substrate concentration in a fed-batch bioreactor. The following model was developed using conventional mass balance equations. These equations were used to build the model in python using several modules including NumPy, Matplotlib, SciPy, and the Bokeh Library for data visualization. This model can easily be used to account for different initial conditions, allowing us to optimize our bioreactor without ever stepping foot in the lab. Chromoprotein Expression Modeling. . Chromoprotein expression modeling was based on the mass balance and expression equations used in the bioreactor modeling above. In a fed-batch bioreactor, the amount of protein expression is heavily influenced by the concentration of substrate in the vat, as well as flow rate. Protein Folding Model. . Predicting the 3D structure of our fused protein part was important in the scope of this project. The csgA parts (⍺ and 𝝲 subunits) are relatively much smaller than the chromoprotein, and we worried that the disparity would cause potential issues in the tertiary structure of the completed protein. In order to predict the 3D structure, we chose to use AlphaFold2, an AI program that would allow us to both visualize our proteins. Using ALphaFold, Cornell iGEM was able to produce the following models of our proteins with relatively high confidence: Laminar Flow Modeling. . Inspired by Cornell iGEM team’s previous project - Collatrix, we have utilized ANSYS to analyze the behavior of hydrogel flow through the syringes that are used for our 3D bioprinter and hand-held bioprinter. The aim of the model was to 1) determine which syringe is best suited for hydrogel extrusion in regards to its capacity and diameter 2) study fluid flow of hydrogel in syringes, which has a higher viscosity than water. Through plotting pressure and velocity contour of the hydrogel, we were able to predict the extrusion of the hydrogel for each 1mL/cc, 3mL/cc, and 20mL/cc syringes. Another inspiration regarding hydrogel modeling using ANSYS was from University of Laval 2021 team, aSAP, where they found a way to homogenize their enzyme with ropy maple syrup by modeling the mixing process of ropy maple syrup in a barrel using ANSYS FLUENT. To do so, team aSAP incorporated ropy maple syrup’s distinctive properties, such as viscosity and density into their ANSYS modeling. Similarly, Micromurals also work with a biomaterial rather unique - hydrogel. As hydrogels exhibit both solid-like and liquid-like properties, specific fluid modeling had to be done in order to determine the compatibility with our syringes. While Micromurals aims to synthesize our own colored hydrogels, the design of the 3D printer and handheld bioprinter needed to be finalized before incorporating the final hydrogel we would be working with. To successfully merge these components, it was necessary to synthesize a gelatin-based hydrogel mimetic for testing purposes. We input the experimental data that we obtained from extrusion testing, such as viscosity, density, and elasticity of the hydrogel into our ANSYS model to study the unique properties of the hydrogels that we synthesized. Given that syringes are cylindrical and flow occurs at a lower speed, 2D laminar flow was calculated using ANSYS FLUENT. Key assumptions made for the modeling were: Hydrogel properties; since we do not have the experimental data of viscosity and density of the hydrogel, we used 2.420 kg/m³ for density, and 0.02 kg/ms for viscosity. X-velocity set to 0.01 m/s as an initial estimate Boundary conditions set set to: inlet-velocity; walls - no slip (velocity is zero at the syringe wall); outlet-pressure After implementing our hydrogel properties and initial x-velocity, we have plotted velocity and pressure contour in each syringe. We observed a coherent trend in the velocity contour of all three syringes, confirming that the velocity would be higher at the syringe tip, and lower at the surface of the syringe due to the boundary conditions set to “non-slip”. This matched our initial estimate of how the velocity would accelerate as the diameter of the syringe decreases towards the syringe tip. We also observed that the velocity builds up more gradually as the diameter of the syringe decreases, as in the 1 mL/CC syringe, the velocity contour was smoother, as velocity started increasing from the syringe body, while 3 mL/CC and 20 mL/CC syringes had a rather sudden acceleration at the syringe tip. In regards to the pressure contour, we observed that the pressure within the syringe was high, and slowly decreased towards the syringe tip as the hydrogel gets released through the tip. For the 20 ml/CC syringe, the static pressure was lower than the 1ml/CC and 3ml/CC syringes, due to an increased volume of the syringe, but was still enough for the hydrogel to be extruded through the tip. Therefore, we have decided that a 1 ml/CC syringe would be best suited for the hand-held bioprinter as we need a more gradual build-up of velocity throughout the syringe, while all 3 types of syringes would be used for our 3D printer. Extrusion. . Extrusion techniques are commonly employed in industries that want to create cutting-edge answers for the more complex needs in the pharmaceutical sectors. Production has typically been done through empirical experience and trial-and-error methods. Based on the ANSYS results, we observed that velocity built gradually in a 1ml syringe and fluid inside stayed rather stable with little acceleration. We aim to physically test both 20 ml and 1 ml to further confirm our ANSYS model. Micromural extrusion modeling and testing is inspired by the Penn iGEM 2019, where they investigated a property of wax drops to characterize the performance of our printer and optimize mechanical and electrical settings for better performance. MicroMurals also focus on 3D printing but with hydrogel which will have a different heating temperature and surface tension while printing. Our project emphasizes on forming continuous and additive printing other than single droplet, which require a more precise and delicate extruder. Our extrusion testing obtained data of printing speed and time from different sizes of syringe to predict the best printing condition for 3D printers. We used the procedure linked here: Extrusion Method Testing Therefore, extrusion testing focuses on flowing and exiting. Flow modeling can be simulated by ANSYS modeling and the exit die was tested by extrusion testing. First, we used Fusion 360 to construct a test frame as shown in the figure below. Syringes of various sizes would fit in the central hole, and weight would be added at the top to provide a steady force. By doing the extrusion testing, we aimed to determine the best syringe size and force applied, as well as the ideal printing speed. After reloading the syringe, we assembled the test frames. We put two known weights on the top as a consistent force to push the syringe and recorded time it took for hydrogel to run out. By experimenting with different ratios of hydrogels, we obtained the data listed below: According to Newton's second law, force = mass multiplied by acceleration. We calculated acceleration by vf^2=2 ax and vf=0+at. Then flow rate is calculated by Mass flow rate formula: Mass flow rate = ρ * Volumetric flow rate = ρ * A * v, where ρ is fluid density. And here flow rate is defined to be the volume of fluid passing by some location through an area during a period of time. By doing all these calculations, we will obtain the following data in the table. We will have the force that applied on the hydrogel plus the printing rate corresponding to this force. Therefore, according to our extrusion testing results, ratios of 1:9, 1:10, 1:11 hydrogels have the properties that work best in 3D printing. Meanwhile, with the same amount of force applied, inside a 20 ml syringe, pressure is lower than 1 ml syringe. This decreased the printing rate and flow rate inside the syringe, which also made 1 ml a better option for printing. After testing gelatin and glutaraldehyde hydrogel, we are trying to enhance the mechanical properties of the gel by combining CD nanofiber into it. Professor Uyar from Cornell Human Ecology Department kindly offered us CDnanofiber from his lab to experiment on. According to Professor Uyar’s instructions we first cut the Nanofiber into pieces and soaked them in warm water to break them down faster. After a day sitting on the bench, we performed sonicating with pulse on 3 seconds, pulse off 4 seconds and repeated for 20 mins to thoroughly mix water with fibers. The process pictures are shown below. Our nanofiber is not breaking down at a speed that we aspired to, and currently we are trying to increase the efficiency. If we successfully made the CD nanofiber hydrogel, we would be testing its properties and perform VOC uptake testing as well.

Bioreactor Modeling

The scope of this project necessitated modeling parameters like cell, product, and substrate concentration in a fed-batch bioreactor. The following model was developed using conventional mass balance equations. Before we describe the model, we first define the following variables:
  • \(X:\) Cell concentration, cells/L
  • \(P:\) Product concentration, g/L
  • \(S:\) Substrate (general) concentration, g/L
  • \(S_f:\) Substrate (feed) concentration, g/L
  • \(F(t):\) Flow rate, L/hr
  • \(V:\) Tank volume, L
  • \(Y_{X/S}:\) Biomass-yield coefficient (the ratio of cell mass to mass of substrate)
  • \(Y_{P/X}:\) Product-yield coefficient (the ratio of product mass to cell mass)
  • \(\mu(S):\) Specific growth rate (function of substrate concentration)
  • \(K_S:\) Half-saturation constant (where specific growth rate is half its maximum)
Using these variables, the following equations were determined to be used in the mass balance:
  • \(\mu(S)=\mu_{max} \frac{S}{K_S+S}\), the Monod Kinetic Model
  • \(r_g(X,S) = \mu(S)X\), rate of production of fresh cell biomass (cell-specific growth.
  • \(r_P(X,S) = Y_{P/X}r_g(X,S)\), the rate of product production
The mass balance equations used in the model were as follows: $$\tag{1} \frac{d(XV)}{dt}=V r_g(X,S)$$ $$\tag{2} \frac{d(PV)}{dt}=V r_P(X,S)$$ $$\tag{3} \frac{d(SV)}{dt}=FS_F-\frac{1}{Y_{X/S}}Vr_g(X,S)$$ $$\tag{4} \frac{dV}{dt}=F(t)$$ Equation 1 describes the change in number of cells with respect to time. Equation 2 describes changes in grams of product with respect to time. Equation 3 describes the change in grams of substrate with respect to time. Finally, equation 4 describes the change in volume in the bioreactor.

Since volume in the bioreactor is not necessarily constant, all of the main parameters (\(X, P, S, V\)) can be affected by dilution. Using differential calculus, we can equations 1-4 together to account for this effect. These final equations can be seen below: $$\tag{1a} \frac{d(X)}{dt}=-\frac{F(t)}{V}X+r_g(X,S)$$ $$\tag{2a} \frac{d(P)}{dt}=-\frac{F(t)}{V}P+r_P(X,S)$$ $$\tag{3a} \frac{d(S)}{dt}=\frac{F(t)}{V}(S_f-S)-\frac{1}{Y_{X/S}}r_g(X,S)$$ $$\tag{4/4a} \frac{dV}{dt}=F(t)$$ These equations were used to build the model in python using several modules including NumPy, Matplotlib, SciPy, and the Bokeh Library for data visualization. The resulting graph is an example of the kind of visualization this model can produce:

Figure 1: Plot of bioreactor model depicting cell mass, product mass, substrate mass, and tank volume changes over time
This model can easily be used to account for different initial conditions, allowing us to optimize our bioreactor without ever stepping foot in the lab.

Chromoprotein Expression Modeling

Chromoprotein expression modeling was based on the mass balance and expression equations used in the bioreactor modeling above. In a fed-batch bioreactor, the amount of protein expression is heavily influenced by the concentration of substrate in the vat, as well as flow rate. $$ \tag{1} r_P(X,S) = Y_{P/X}r_g(X,S) $$ Equation 1 above describes the rate of protein production, dependent on \(Y_{P/X}\), the product yield coefficient, and the function \(r_g(X,S)\), which is described below: $$r_g(X,S) = \mu(S)X $$ Where \(\mu(S)\) is the monod kinetic model function as described earlier, and \(X\) is the cell concentration. The above equations were fed into the differential equation: $$ \frac{d(P)}{dt}=-\frac{F(t)}{V}P+r_P(X,S) $$ Where \(\frac{F(t)}{V}\) is the flow rate divided by the volume of the bioreactor, and \(P\) is the chromoprotein concentration. \(r_P(X,S)\) is as described above. This differential equation was solved using the SciPy python package and the odeint() function. Below is a table showing the parameters that were used in our model:
Variable Value
YP/X **
𝜇(S) 1.07 hr-1 [1]
F(t) 25.2 L/hr
V 1.75 L
** Variable not available in literature

Table 1. Parameters list for expression modeling variables

Below is a graph produced with the modeling equations and parameters shown above. Unfortunately, because the protein we are trying to produce is a novel fusion protein, we were not able to determine a product yield coefficient parameter from existing literature. This is one of the important drawbacks of our model.

Figure 2: Chromoprotein expression plot over time

There are many ways in which we can improve our model and general modeling practice given enough time. One of the ways in which Cornell iGEM thought to improve our model was to add in differential equations for transcriptional and translational control of our system. The following differential equation may be used to model transcription [2].This could then be fed into our model to more accurately predict the amount of protein produced:

$$ \frac{dmRNA}{dt}=f(p)-Vr $$

Where:

  • \(f(p)\), a linear transcription model. This can be a number of different linear functions, but some options include combining the effects of activators or inhibitors. In our case, our construct is induced with the presence of IPTG, so that could be factored in.
  • \(V\), the degradation rate of mRNA.
  • \(r\), the concentration of mRNA.

The below equation is the translational control equation, which takes input from the transcription differential equation:

$$ \frac{dp}{dt}=Lr-Up $$

Where:

  • \(L\) are translational constants. These could be as simple as translation rate, or take into account more complicated factors, such as translation efficiencies, product yield coefficients, etc.
  • \(r\), the concentration of mRNA from the equation above.
  • \(U\), protein degradation rates.
  • \(p\), the protein concentration.
The values for the above parameters can be either found or reasonably estimated through literature review and using resources such as https://bionumbers.hms.harvard.edu/search.aspx.

Protein Folding Simulation

Predicting the 3D structure of our fusion protein part was important in the scope of this project. The csgA parts (⍺ and γ subunits) are relatively smaller than the chromoprotein, and we worried that the disparity would cause potential issues in the tertiary structure of the completed protein.

In order to predict the 3D structure, we chose to use AlphaFold2, an AI program that would allow us to visualize our proteins [4,5]. Using AlphaFold, Cornell iGEM was able to produce the following models of our proteins with relatively high confidence:

csgA⍺ - Blue

csgA⍺ - Pink

csgA⍺ - Lime

csgA⍺

All of our models compiled in AlphaFold 2.0 successfully. From the models themselves, there did not appear to be severe misfolding. As we had hoped, it seems that the domains of each part of the fusion protein are separate and fused with a linker. This suggests that our final constructs will still feasibly be able to show color despite being fusion proteins.

Laminar Flow Modeling

Inspired by Cornell iGEM team’s previous project - Collatrix, we have utilized ANSYS to analyze the behavior of hydrogel flow through the syringes that are used for our 3D bioprinter and hand-held bioprinter. The aim of the model was to 1) determine which syringe is best suited for hydrogel extrusion in regards to its capacity and diameter 2) study fluid flow of hydrogel in syringes, which has a higher viscosity than water. Through plotting pressure and velocity contour of the hydrogel, we were able to predict the extrusion of the hydrogel for each 1mL/cc, 3mL/cc, and 20mL/cc syringes.

Another inspiration regarding hydrogel modeling using ANSYS was from University of Laval 2021 team, aSAP, where they found a way to homogenize their enzyme with ropy maple syrup by modeling the mixing process of ropy maple syrup in a barrel using ANSYS FLUENT. To do so, team aSAP incorporated ropy maple syrup’s distinctive properties, such as viscosity and density into their ANSYS modeling. 

Similarly, MicroMurals also work with a biomaterial rather unique - hydrogel. As hydrogels exhibit both solid-like and liquid-like properties, specific fluid modeling had to be done in order to determine the compatibility with our syringes. While MicroMurals aims to synthesize our own colored hydrogels, the design of the 3D printer and handheld bioprinter needed to be finalized before incorporating the final hydrogel we would be working with. To successfully merge these components, it was necessary to synthesize a gelatin-based hydrogel mimetic for testing purposes. We input the experimental data that we obtained from extrusion testing, such as viscosity, density, and elasticity of the hydrogel into our Ansys model to study the unique properties of the hydrogels that we synthesized.

Given that syringes are cylindrical and flow occurs at a lower speed, 2D laminar flow was calculated using ANSYS FLUENT. Key assumptions made for the modeling were:

  1. Hydrogel properties; since we do not have the experimental data of viscosity and density of the hydrogel, we used 2.420 kg/m³ for density, and 0.02 kg/ms for viscosity.
  2. X-velocity set to 0.01 m/s as an initial estimate
  3. Boundary conditions set set to: inlet-velocity; walls - no slip (velocity is zero at the syringe wall); outlet-pressure

After implementing our hydrogel properties and initial x-velocity, we have plotted velocity and pressure contour in each syringe. 

Figure 3: Contours of Velocity Magnitude [m/s], 1mL/CC syringe
Figure 4: Contours of Static Pressure [Pa], 1mL/CC syringe
Figure 5: Contours of Velocity Magnitude [m/s], 3mL/CC syringe
Figure 6: Contours of Static Pressure [Pa], 3mL/CC syringe
Figure 7: Contours of Velocity Magnitude [m/s], 20mL/CC syringe
Figure 8: Contours of Static Pressure [Pa], 20mL/CC syringe

We observed a coherent trend in the velocity contour of all three syringes, confirming that the velocity would be higher at the syringe tip, and lower at the surface of the syringe due to the boundary conditions set to “non-slip.” This matched our initial estimate of how the velocity would accelerate as the diameter of the syringe decreases towards the syringe tip. We also observed that the velocity builds up more gradually as the diameter of the syringe decreases, as in the 1 mL/CC syringe, the velocity contour was smoother, as velocity started increasing from the syringe body, while 3 mL/CC and 20 mL/CC syringes had a rather sudden acceleration at the syringe tip. In regards to the pressure contour, we observed that the pressure within the syringe was high, and slowly decreased towards the syringe tip as the hydrogel gets released through the tip. For the 20 ml/CC syringe, the static pressure was lower than the 1ml/CC and 3ml/CC syringes, due to an increased volume of the syringe, but was still enough for the hydrogel to be extruded through the tip. Therefore, we have decided that a 1 ml/CC syringe would be best suited for the hand-held bioprinter as we need a more gradual build-up of velocity throughout the syringe, while all 3 types of syringes would be used for our 3D printer.

Extrusion

Extrusion techniques are commonly employed in industries that want to create cutting-edge answers for the more complex needs in the pharmaceutical sectors. Production has typically been done through empirical experience and trial-and-error methods. Based on the ANSYS results, we observed that velocity built gradually in a 1ml syringe and fluid inside stayed rather stable with little acceleration. We aim to physically test both 20 ml and 1 ml to further confirm our ANSYS model.

Hydrogel extrusion modeling and testing is inspired by the Penn iGEM 2019, where they investigated a property of wax drops to characterize the performance of our printer and optimize mechanical and electrical settings for better performance. MicroMurals also focus on 3D printing but with hydrogel which will have a different heating temperature and surface tension while printing. Our project emphasizes on forming continuous and additive printing other than single droplet, which require a  more precise and delicate extruder. Our extrusion testing obtained data of printing speed and time from different sizes of syringe to predict the best printing condition for 3D printers. We used the procedure linked here: Extrusion Method Testing

Therefore, extrusion testing focuses on flowing and exiting. Flow modeling can be simulated by ANSYS modeling and the exit die was tested by extrusion testing. First, we used Fusion 360 to construct a test frame as shown in the figure below. Syringes of various sizes would fit in the central hole, and weight would be added at the top to provide a steady force.

Figure 9. Fusion 360 model of the Test Frame
Figure 10: Setup of Test Frame
By doing the extrusion testing, we aimed to determine the best syringe size and force applied, as well as the ideal printing speed. After reloading the syringe, we assembled the test frames. We put two known weights on the top as a consistent force to push the syringe and recorded time it took for hydrogel to run out. By experimenting with different ratios of hydrogels, we obtained the data listed below:

No.Trial Hydrogel Ratio Syringe size(ml) Hydrogel Mass (g) Weight Mass (g) Density (g/ml) Displacement (cm) Time(sec)
1 1:10 20 1.391 247.855 1.198 Failed to extrude
2 1:05 20 1.916 583.855 1.00 0.9 40
3 1:08 20 1.797 2300 1.0593 0.95 28.83
5 1:09 20 2.097 2300 1.0593 0.7 27.95
6 1:11 20 2.09 2300 1.5145 0.7 3.46
7 1:10 1+tip 0.235 2300 1.198 1.2 14.7
8 1:10 1+goodtip 0.41 1000 1.198 2.9 96.67
Table 2: Measurement from testing hydrogel with 20 ml 1 ml syringes

We assume a 1D steady laminar poiseuille flow:
\(Q = \frac{\pi Pr^4}{8 \eta l}\); where \(Q\) is the volume flow rate. \(P\) is the applied pressure, \(r\) is the radius, \(\eta\) is the viscosity, and \(l\) is the length of the syringe (12.7 mm). We assume that the needle length is the dominant cause of flow restriction.
Then flow rate is also:
\(Q = \frac{\pi Pr^4} {8 \eta l} = r^2 v\), where \(\rho\) is fluid density and \(v\) is the flow velocity.
And here, flow rate is defined to be the volume of fluid passing by some location through an area during a period of time. By doing all these calculations, we get the viscosity by rearranging.
\(\eta = \frac{Pr^2}{8vl}\); where pressure is caused by the weight: \(P = \frac{F}{A} = \frac{Mg}{r^2}\) \(\eta= \frac{Mg}{8vl}\)

No.Trial Velocity (mm/sec) A(mm/sec^2) Flow Rate (ml/sec) Force (N)
1 Failed
2 0.45 0.01125 0.05 0.006568
3 0.659035727 0.022859373 0.069372182 0.052577
5 0.500894454 0.01792109 0.071556351 0.041219
6 4.046242775 1.169434328 0.578034682 2.689699
7 1.632653061 0.111064834 0.136054422 0.255449
8 0.599979311 0.006206469 0.020688942 0.006206
Table 3: Velocity and printing speed from testing hydrogel with 20 ml 1 ml syringes

From the data above, we can see surface tension effects at the end of the needle, and viscoelastic effects dominate. The viscosity is much higher than expected in our fluid simulations, thus the hydrogel behaves more like a solid. The velocity and acceleration is within an adequate range between 1-3 mm/sec. Meanwhile, 1:10 and 1:11 ratios of hydrogel seem to have great consistency in adding layers. The image below illustrates when we experimented with the 1:10 ratio hydrogel. The hydrogel successfully mixed together and was able to perform additive printing.

Figure 11: Extrusion Testing of Hydrogel additive property

Therefore, according to our extrusion testing results, ratios of 1:9, 1:10, 1:11 hydrogels have the properties that work best in 3D printing. Meanwhile, with the same amount of force applied, inside a 20 mL syringe, pressure is lower than 1 mL syringe. This decreased the printing rate and flow rate inside the syringe, which also made 1 mL a better option for printing. 

Cyclodextrin (CD) Nanofiber Incorporation

After testing gelatin and glutaraldehyde hydrogel, we attempted to enhance the mechanical properties of the gel by combining CD nanofiber into it, which have also been used for VOC-uptake applications [11]. Professor Tamer Uyar from Cornell Human Ecology Department kindly offered us 0.6 g of electrospun CD nanofiber from his lab to experiment on.

Figure 12: CD nanofiber in test tube (left) sample in sonicating machine (right).

According to Professor Uyar’s instructions and previous sources [12] we first cut the nanofiber into very small pieces and soaked them in warm water to break them down faster. After a day sitting on the bench, we used a magnetic stirrer to stir them for some time until they were mostly hydrated and were floating in the tubes. We performed sonication with pulse on 3 seconds, pulse off 4 seconds and repeated for 20 mins to further disperse them in water. The process pictures are shown below. Our nanofiber is not breaking down at a speed that we hoped, and currently we are trying to increase the efficiency. Here is the CD Nanofiber Procedure for our experiment.

Figure 13: Post Sonication- Tube A(left) original size Tube B(right) chopped nanofiber sheet into pieces before sonication.
If we successfully made the CD nanofiber hydrogel, we would be testing its properties and perform VOC uptake testing as well.

Sources

[1] Selvarasu, S., Ow, D. S.-W., Lee, S. Y., Lee, M. M., Oh, S. K.-W., Karimi, I. A., & Lee, D.-Y. (2009). Characterizing Escherichia coli DH5alpha growth and metabolism in a complex medium using genome-scale flux analysis. Biotechnology and Bioengineering, 102(3), 923–934. https://doi.org/10.1002/bit.22119

[2] Chen, T., & He, H. L. (n.d.). MODELING GENE EXPRESSION WITH DIFFERENTIAL EQUATIONS a. 12.

[3] Bionumbers - the database of useful biological numbers. (n.d.). Retrieved from https://bionumbers.hms.harvard.edu/search.aspx.

[4] AlphaFold Protein Structure Database: Massively expanding the structural coverage of protein-sequence space with high-accuracy models | Nucleic Acids Research | Oxford Academic. (n.d.). Retrieved October 12, 2022, from https://academic.oup.com/nar/article/50/D1/D439/6430488?login=false.

[5] Jumper, J., Evans, R., Pritzel, A., Green, T., Figurnov, M., Ronneberger, O., Tunyasuvunakool, K., Bates, R., Žídek, A., Potapenko, A., Bridgland, A., Meyer, C., Kohl, S. A. A., Ballard, A. J., Cowie, A., Romera-Paredes, B., Nikolov, S., Jain, R., Adler, J., … Hassabis, D. (2021). Highly accurate protein structure prediction with AlphaFold. Nature, 596(7873), Article 7873. https://doi.org/10.1038/s41586-021-03819-2.

[6] Model. Cornell iGEM 2021 - Collatrix. (n.d.). Retrieved from https://2021.igem.org/Team:Cornell/Model.

[7] ULaval iGEM 2021 - aSAP. (n.d.). Retrieved from https://2021.igem.org/Team:ULaval.

[8] Penn iGEM 2019. (2019, October). Results Characterizing our device. 2019.igem.org: Team Penn. https://2019.igem.org/Team:Penn/Results.

[9] S.A.Tabatabaei. (2015, September). 2022, Analytical modeling of the extrusion process using the electrostatics concept. sciencedirect.com. https://www.sciencedirect.com/science/article/pii/S0167663615000769.

[10] Marko Hyvärinen . (2020, December 6). The Modeling of Extrusion Processes for Polymers—A Review. Mdpi.com:polymers. https://www.mdpi.com/2073-4360/12/6/1306.

[11] Celebioglu, A., Sen, H. S., Durgun, E., & Uyar, T. (2016). Molecular entrapment of volatile organic compounds (VOCs) by electrospun cyclodextrin nanofibers. Chemosphere, 144, 736-744.

[12] Yu Huang, Xiufang Li, Zhentan Lu, Huan Zhang, Jiangxi Huang, Kun Yana and DongWang a. (2020). Nanofiber-reinforced bulk hydrogel: preparation and structural, mechanical, and biological properties. ROYAL SOCIETY OFCHEMISTRY. https://pubs.rsc.org/en/content/articlelanding/2020/tb/d0tb01948h#!divAbstract.