Journal Entries

July 22 - July 26

Weekly Summary

This week was dedicated to troubleshooting results and protocols. The team started ethanol and salt precipitating DNA after the digests with double the amount of DNA. A wash step was also added to the protocols after precipitation, with the vacufuge step reduced to 15 minutes maximum. Additionally, SOC protocols were added for use in place of standard LB for transformations. Each PCR step was adjusted to increase activity and specificity based on our constructs and ingredients. Since DNA amounts were doubled for digests, the team did a PCR spree on the full constructs. More pSBIA3 backbone was digested since those last week were not, causing the failed transformations. We also talked to Dr. John Baker to discuss the SynBio discussion that will be organized in September.

Figure 1.1% agarose gel showing colony PCR of standard and pelleted SGR and CBR. Gel ran at 100V for 30 min. Expected lengths of bands are 9000 bp for SGR and 9999 for CBR. Row 1 : Lane 1: Ladder - NEB 1kb+ | Lanes 2-4: Standard SGR| Lanes 6-12: Pelleted SGR. Row 2: Lane 1: Ladder -NEB 1kb+ | Lane 2-5 : Standard CBR | Lane 6-13: Pelleted CBR.

September 9 — September 15

Weekly Summary

More transformations and cPCR were done for SGR. The first thin layer chromatography experiment was run using flexible cellulose TLC plates with 100% hexane as solvent. However, the first plate did not run well because the chamber was not saturated enough. We were unsure of what the orange color seen on the solvent fron on top of chlorophyll #3 and acidified chlorophyll #3 and #2 were. This might be due to pheophytin or chlorophyll all the way to the solvent front. Chris acidified chlorophyll sample to create pheophytin, and absorbance values corresponding to chlorophyll wavelengths decrease and pheophytin wavelengths increased. Pheophorbide did not run on the cellulose plates with 100% hexane. Team also attempted to use the big TLC chamber and run cellulose sheets on 100% hexane but the chamber was not saturated enough so the solvent kept evaporating from the plate. Another SDS-PAGE gel was run, and showed fourth confirmation of PPH and third confirmation of HCAR purification. Pheophorbide a was solubilized in 25% acetone. Dark treatments with pheophorbide paper disc tests across different concentrations on fungal cultures were started. More fungal cultures for light treatments (1400 lumens, white LED) were made with double disks at 1.5 cm away from the center of the fungi to the disk center, with concentrations 0mg/ml, 5 mg/ml, 25mg/ml, and 35mg/ml. The double disks were done to account for the possibility of pheophorbide diffusion into the agar plates.

Figure 1. First TLC. FOURTH PROTEIN GEL. 4/4 Confirmation of PPH and 3/4 Confirmation of HCAR. Samples run were from September 4, 2019 Column purifications. OLD BL21 was from September 14, 2019 (?).

Figure 2. SDS-page gel caption. FOURTH PROTEIN GEL. 4/4 Confirmation of PPH and 3/4 Confirmation of HCAR. Samples run were from September 4, 2019 Column purifications. OLD BL21 was from September 14, 2019 (?).

Figure 3. Pheophorbide a, pheophytin, and chlorophyll visualizaiton. Samples were eluted by Thin Layer Chromatography using 80:20 acetone: hexane, and methanol as solvent on flexible cellulose plates.

Figure 4. Pheophorbide fungal disc testing layout. Samples were eluted by Thin Layer Chromatography using 80:20 acetone: hexane, and methanol as solvent on flexible cellulose plates.

Figure 5. Fungi day 2. Samples were eluted by Thin Layer Chromatography using 80:20 acetone: hexane, and methanol as solvent on flexible cellulose plates.

Figure 6. Fungi day 3. Samples were eluted by Thin Layer Chromatography using 80:20 acetone: hexane, and methanol as solvent on flexible cellulose plates.

October 7 — October 13

Weekly Summary

Team did more digests, ligations, transformations, cPCR, and minipreps for SGR and CBR. pSBIA3, PPH, and HCAR proteins were ammonium sulfate precipitated, but the salt was not fully dissolved so the supernatants from the three elutions were aggregated in a falcon tube. We were unsure wether the proteins would still be functional, but ran the solutions on silica plates using methanol as a solvent. Since pheophorbide and pheophytin were still not easily distinguishable from each other, a solvent mixture of 70:30 methanol:hexane was used. More subcultures were made, induced, protein-purified, and ammonium-sulfate precipitated. New pheophytin stock was made using chlorophyll from fresh spinach, and neutralized to pH 8.0 to ensure pheophytinase will not denature.

Figure 1.1% agarose gel showing colony PCR of standard and pelleted SGR and CBR. Gel ran at 100V for 30 min. Expected lengths of bands are 9000 bp for SGR and 9999 for CBR. Row 1 : Lane 1: Ladder - NEB 1kb+ | Lanes 2-4: Standard SGR| Lanes 6-12: Pelleted SGR. Row 2: Lane 1: Ladder -NEB 1kb+ | Lane 2-5 : Standard CBR | Lane 6-13: Pelleted CBR.

Figure 2. Pheophytin sample visualization. Samples were eluted by Thin Layer Chromatography using methanol:hexane (70:30) solvent on a silica plate.

Figure 3.Old pheophytinase elution fraction (non-ammonium sulfate precipitated) reaction, pheophytin (acidified chlorophyll stock) and pheophorbide a visualization. Samples were eluted by Thin Layer Chromatography using methanol:hexane (70:30) solvent on a silica plate.

Figure 4. Failed ammonium sulfate-precipitated pheophytinase reaction, pheophytin, and pheophorbide a visualization. Samples were eluted by Thin Layer Chromatography using methanol:hexane (70:30) solvent on a silica plate.

Figure 5. Pheophytin, pheophorbide a, chlorophyll a+ b visualization. Samples were eluted by Thin Layer Chromatography using methanol:hexane (70:30) solvent on a silica plate.

May 6 — May 10

Weekly Summary

We had the task of modelling chlorophyll degradation to its derivatives allowing us to know the respective concentrations over time and which pathways will be more favored in the process. This is useful for knowing which enzymes to overexpress to obtain the desired derivative after chlorophyll a and b have been extracted from the canola oil.

For a separate model, we were given the task to simulate an environment with a chlorophyll binding protein. The end goal is to feed homologies from the protein previously discovered (called 6GIX) into our simulations and measure the respective fitnesses (assuming 6GIX is incompetent of binding to ligands dissolved in oil). Fitness would be measured by its binding affinity to chlorophyll that is dissolved in the surrounding oil and its ability to maintain a steady state (resist denaturation). So we hit the ground running with GROMACS - a computational chemistry software that simulates molecule interactions with great detail. It is of vital importance that we are given access to ARC - the supercomputer hosted at the University of Calgary which will drastically speed the times of our simulations. We started using a Linux machine to install the massive software - we chose the "quick and dirty" installation (which still took 1 hour).

We started our first simulation with the goal of observing our protein in water for one nanosecond (the basic tutorial was followed here: http://www.mdtutorials.com/gmx/lysozyme/index.html). The environment was captured in a box the size of 6GIX with only a 1nm gap measured off the maxima of 6GIX in all directions. From there we placed the protein topology with the four chlorophyll ligands removed as a .pdb file and added a respective amount of solvate (water). 6GIX has a -10e charge without its ligands so to ensure charge balance 10 sodium ions were added to the scene. These steps did not take long to produce for our linux desktop however the next steps of getting the protein to its minimum energy state and creating equilibrium for temperature and pressure took around three days to simulate. The last step of the simulation which would produce molecular dynamics of the system for one nanosecond was unable to finish before we decided to rerun this simulation of ARC.

While ARC was processing out job we looked into instaling VMD (Visual Molecular Dynamics), a software that takes the simulation output from GROMACS and creates a graphical representation of it. From here we can visualize molecular positions in timesteps less than a nanosecond and make use of scripting to measure fitnesses of the protein - how this measurement will be done is still being looked into. We discussed our next simulations and decided on a few intermediary models before we simulate the most crucial one that will be most applicable to our project. We would be wasting computing if we went straight to the most advanced simulation without knowing whether the protien in question can even be partially in oil in the first place (let alone take chlorophyll out of oil). Our next intermediary simulation will be immersing the protein in oleic acid (which composes 61% of canola oil) to see how GROMACS simulates and handles denaturation (6GIX is insoluble in lipids).

Chlorophyll degradation kinetic models began to be developed. Literature shows that only a first order rate law mechanism has been proposed for chlorophyll degradation studies. This model has been previously use to characterize degradation for in-vivo experiments in coleslaw, olives, pickles, and other foodstuff. Model pathway considers 2 parallel pathways, (1. chlorophyll -> chlorophyllide -> pheophorbide) & (2. chlorophyll -> pheophytin -> pheophorbide). No particullar consideration was given to the enzymes or any of the "hydroxymethyl" intermediates in the chlorophyllide B pathway. Given explanation for this particular model seems to be that A & B pathways share the chlorophyllase and Mg-dechalatase enzymes, so the chemical groups can be lumped into one (model does not consider cofactor concentration, cycling, or conversion between chlorophyll A and B). Experimental data from papers seems to fit model fairly well, however this may be due to large experiment periods. Matlab scripts for the first order rate law were written to hopefully measure the validity of the model with in-vitro data.

May 13 — May 17

Weekly Summary

Work on chlorophyll degradation models continued. Continued literature review did not reveal any new kinetic mechanisms in the suggested pathway - however a small amount of studies do outline thermal and pH effects, as well as propose models for this. Possible reasons that new mechanisms have not/rarely been considered is because of natural species concentration in vivo, long reaction time, and a lack of relevance in industrial processing. Mechanisms have to be proposed then, for each intermediate step and parameters have to be obtained from experimental results from best fit data.

Enzymes between each step began to be studies, to determine the best potential mechanism. Currently, michaelis-menten model would seem to be a possibility for degradation of both chlorophyll A and B to chlorophylide A, however it does not take into consideration selectivity for either - (paper shows that the B pathway is selectively favorable for some reason). Other potential mechanism is ping-pong, but there is no evidence to suggest that the enzyme is tranformed between complexes for any reason (yet). Ping-pong might be a good choice for any of the "mirror-steps" but, there is no evidence to suggest enzyme transformation - more reading into mechanisms must be considered.

Chlorophyll degradation subgroup and model workflow were redefined. Instead of looking to optimize parts sequence to maximize amount of each enzyme produced, enviornmental conditions should be optimized first. Initially, kinetic data will be produced for experiments between each step (chlorophyll B to chlorophyllide B via chlorophyllase enzyme). Once trials are performed and kinetic parameter, experiments with the entire systme will be done at once - so cofactor concnetrations, pH, and temperature will be optimised for this experiment with the kinetic models. An inducible expression model defining the rates of production for each enzyme can also be built, but was deemed to be possibly too dificult, and not a priority. Scirpting work was continued, with an emphasis on the model framework itself. Essentially, enzyme mechanisms, parallel rate law systems, fitting and optimization algorithms are being structured independently so when experimental data is available, model framework is set up and ready to go.

May 20 — May 24

Weekly Summary

Protein modelling:

We finished the simulation of 6GIX solvated in oil and GROMACS blew up, giving us fatal errors due to too large of molecule movement. This is not anything interesting because we thought that 6GIX would denature in oil. We found packmol which is a program that offers a more intuitive interface for setting up MD (Molecular Dynamic) environments to use in GROMACS. Packmol takes some time to export these environments, and currently we are building an environment where 6GIX is half solvated in oleic acid and half solvated in water. Another environment (which is the most analogous to the applicability of our project idea) is being built as well, however we are struggling getting the data structure for chlorophyll and the number of molecules needed within the area specified that would constitute a liquid phase. This model would encapsulate 6GIX in a sphere of water with oil outside of this sphere which would contain dissolved chlorophyll. It is in this simulation where we would hope to see chlorophyll move across the surfactant layer and bind with 6GIX in the water. We also are looking to measuring the fitness of 6GIX by evaluating the free binding affinity which can presumably be measured computationally. The main issues looking forward are how the binding affinity is affected for later ligands after one ligand attatches - the optimal number of ligands 6GIX can contain is unknown at this point. Another issue is simulating 6GIX with a histone tag, which is needed for the protein extraction after E.Coli produces its complex. To add a histone tag to a .pdb file might best be accomplished using the software pymol.

Degradation Modelling:

Degradation pathway was redefined to sequential steps: Chlorophyll B -> 7-hydroxymethyl Chlorophyll a -> Chlorophyll a -> pheophytin a -> pheophorbide a. Respective enzymes for each reaction are: Chlorophyll b reductase, 7-hydroxymethyl chlorophyll a reductase -> Mg-dechalatase -> pheophytinase.

New workflow: continue to set up scripts indipendent of experiments -> using MM mechanism for all, will be developing ratio tests for the 2-step assays - have not looked at temperature, pH, or cofactor dependence yet. For temp, initially will be using arrhenius correlation - some literature has shown adding a cp*pH term fits pH, but this is done on fruit over large periods of time - dependent deviations may not be measurable.

Circuit optimization

Began learning the focused circuit construct. Using previous years' work as a learning platform.

Packmol finished the build for 6GIX in the centre of a surfactant layer of oil and water. On the oil side, 6GIX was completely disintegrated and there was one subunit remaining on the water side, however it looked very denatured. This tells us 6GIX cannot even remain partially in oil. The next environment will be 6GIX immersed in water with a surfactant layer with oil on one side of the cube. This is to see if the surfactant between the oil and water will have an effect on the stability of 6GIX.

June 10 - June 14

Weekly Summary

Protein Modelling

Finished Oleic acid model this week now it is just working with energy minimization in biphasic systems. Learned commands and theory around accounting for periodic boundary conditions. Completed electrostatic interaction modelling and homology structural prediction for 5DQR. This was then utilized to build the tag they will be using.

Italian Wedding Soup

Marinara:

*New project started - spreadsheet analysis of existing market , will include sensitivity analysis to inform minimum effectiveness of separation system to net 0

Pickles/Coleslaw:

Contnued work on temp & pH effect derivations

Meeting with degradation subgroup: Use ordered double substrate mechanisms and random release for reactions (reductases use substrate second, cofactor first) (mg dech & phytinase use substrate first, cofactor 2nd)

Subgroup is working on determining possible release (ask LV).

There is a chemical that will be used for cycling - micha will send papers

Finished double variation scripts - surface plots can be obtained now.

Need to figure out how to set up scripts to work on 2 step, not total cumulative

Sunny Days

PCA and Neural nets are under way and undergoing optimization

Explored RNN's for multivariate regression on time series.

ketCHUp

Monaco

Monaco, after intense work, figuring out dozens of differential equations and multitudes of parameters, was deemed nearly impossibly difficult.

can only be pursued effectively if wet lab does experiments fast and has quantifiable results by August 1st.

will be doing literature review until further notice - possible feedback regulation mechanism could be interesting to increase productive lifetime

SYNBAD is deemed an awful tool to run these on.

Salad Dressings:

Identified polderman equation as possible set up for settler design

Henschke's method could be useful, but it does not include surfactant mechanics AND needs specialised equipment

Started reading into surface response method (a stochastic approach) for separation kinetics - need to read more - relatively simple but need viscometer and electrophoresis thing

Met with Dr. Nashat Nassar - agreed that LLE experiments would be good to perform

Need to build ternary (maybe quartary?)phase diagram

Mentioned either McCabe-thiel or hunter-nash method for analysis would be appropriate (didn't say anything about kremser)

Modelling approach could be to consider emulsion as a homogenous solvent, build ternary phase diagram using optimized formulation from surface response (NEEDS MORE RESEARCH)

For formulation experiments, phase diagrams might need to be constructed for the emulsion composition (could be an alternative to surface response)

Equilibrium of surfactants can be described by winsor types (1,2,3,4). There seem to be different kinetic formulations for each type. Need to consider the transition between winsor types (might be infered from phase diagram - tricritical point seems to be conditions for winsor type 3).

July 2 - July 5

Weekly Summary

Protein Modelling:

Started collecting instantaneous root mean square fluxuation data for 500 discrete time points for a single 6gix monomer. Began generating a massive model to demonstrate the aggregative properties of the 6gix monomers. The initial attempt failed after 80 hours of computation. This was then trouble shooted and an aggregation model was established for both 1 and 5 ns time intervals totalling almost an entire week of computation time.

Italian Wedding Soup/Marinara:

Found paper that outlines economics of farmers

Scale-up:

Created McCabe-Thiele Script for Liquid-Liquid Extractions at low concentrations. Will derive a kremser equation as well to analytically determine stage number. Goal is to consider alternative process operations to syn-bio for comparative assesment (we want syn-bio approach to be clearly better). <<<>>>

More practice with emulsions. testing different formulations and observing size (emulsion structure). Began to look into possible neural net approach - different types of hybrid nets are used in similar studies, but existing approaches have illogical mehtods - like using 1 net to predict different ampiphile effects, and not including HBL or OHBLD measurement. Jake, Rylan, and Andrew mentioned that a classification approach using SVM and/or k-means clustering might be able to interporlate rather than using a neural net.

Began to look into clay filtration - clay bleacing via slurry tank seems to be industry standard. Identified Montromillonite k10 to be a good adsorbent from literature. However, kinetic and equilibrium models for this clay (for chlorophyll extraction) is not immediatly obvious. Needs more research. Need this information to build clay model (comparative assesment).

Met with Dr. Hector de la hoz sigler - chemical kinetics and bioprocess professor. Said that he thinks that current degradation scheme is not scalable. Too many enzymes, and will likely run into cycling issues with specificity of sodium dithionate. I asked him about my immmobilization idea: if enzymes operate as classical heterogenous catalyst, they may likely be re-used easily. Convert chlorophyll into something more aquoues-soluble or something benficial to oil pigmentation or stability. Hector said that you run into the same problem wth recycling regardless. Also gave advice on assays for double-substrate enzymes and method of analysis (hold 1 var constant, obtain K1, do the same to other, obtain K2). Lineweaver burk plot used to be used for obtianing Km, but it is better to do non-linear regression (Sum of squares etc..). Reccomended "Bioprocess Engineering - basic concepts" book.

Sunny Days:

Continued editing of the paper

July 15 - July 19

Weekly Summary

Protein Modelling:

Completed Genetic Algorithm to generate 6gix modifications based on in silico targeted selection. This was used to find suitable replacements to the 12 amino acids that attributed the most variance in the dynamics of the protein as found in the Expectation Maximization clustering based on the functional principal component analysis results.

Economic Modelling:

Found that the cost of direct synthesis is 160 000x the worlds money supply/tonne protein Found the cost of growing the cells in terms of reagents

Scale-up:

Started writing scripts for adsorption column. Other adsorbents like silica and zeolites would be operated using a packed column - would have to include TSA or PSA to operate adsorption process continously. Began to look into these models on aspen adsorption. Also, began to read more into adsopriton isotherms and order montmorillonite clay from sigma.

Met with Dr. Jinghuang Hu - biochemical engineering and bioprocesses professor. Mentioned same idea as hector - current degradation pathway is not cost effecient. Mentioned a possible immobilization technique using lignin nanoparticles instead of hydrogel or fillament set up - forwarded the paper to seb. Immobilize enzymes on nanoparticle surface, then can dilute into a more soluble solvent (similar to previous idea). The study employs this approach for butanol butyruc acidm and butyl butyrate. Dr. Hu offered any resources he can provide us with are at our disposal. Will look more into nanoparticle immobilization.

Phase Diagrams:

Span 80 and tween 80 came in! Also created a dilution phase diagram so we know how much water to add to be able to run many tests, this tool will be used to run experiments. Began to set up emulsion tests with this composition AND tested a quick and dirty way to simulate green oil. Added crude chl-etoh stock to oil in a measured approach to get ~30ppm of chl, and boiled off ethanol. Seemed to be relatively successful.

1 2 95.00 8.66 90.00 0.00 10.00 3 61.67 8.66 56.67 33.33 10.00 4 45.00 8.66 40.00 50.00 10.00 90.00 17.32 80.00 0.00 20.00 5 35.00 8.66 30.00 60.00 10.00 70.00 17.32 60.00 20.00 20.00 6 28.33 8.66 23.33 66.67 10.00 56.67 17.32 46.67 33.33 20.00 85.00 25.98 70.00 0.00 30.00 7 23.57 8.66 18.57 71.43 10.00 47.14 17.32 37.14 42.86 20.00 70.71 25.98 55.71 14.29 30.00 8 20.00 8.66 15.00 75.00 10.00 40.00 17.32 30.00 50.00 20.00 60.00 25.98 45.00 25.00 30.00 80.00 34.64 60.00 0.00 40.00 9 17.22 8.66 12.22 77.78 10.00 34.44 17.32 24.44 55.56 20.00 51.67 25.98 36.67 33.33 30.00 68.89 34.64 48.89 11.11 40.00 10 15.00 8.66 10.00 80.00 10.00 30.00 17.32 20.00 60.00 20.00 45.00 25.98 30.00 40.00 30.00 60.00 34.64 40.00 20.00 40.00 75.00 43.30 50.00 0.00 50.00 11 13.18 8.66 8.18 81.82 10.00 26.36 17.32 16.36 63.64 20.00 39.55 25.98 24.55 45.45 30.00 52.73 34.64 32.73 27.27 40.00 65.91 43.30 40.91 9.09 50.00 12 11.67 8.66 6.67 83.33 10.00 23.33 17.32 13.33 66.67 20.00 35.00 25.98 20.00 50.00 30.00 46.67 34.64 26.67 33.33 40.00 58.33 43.30 33.33 16.67 50.00 70.00 51.96 40.00 0.00 60.00 13 10.38 8.66 5.38 84.62 10.00 20.77 17.32 10.77 69.23 20.00 31.15 25.98 16.15 53.85 30.00 41.54 34.64 21.54 38.46 40.00 51.92 43.30 26.92 23.08 50.00 62.31 51.96 32.31 7.69 60.00 14 9.29 8.66 4.29 85.71 10.00 18.57 17.32 8.57 71.43 20.00 27.86 25.98 12.86 57.14 30.00 37.14 34.64 17.14 42.86 40.00 46.43 43.30 21.43 28.57 50.00 55.71 51.96 25.71 14.29 60.00 65.00 60.62 30.00 0.00 70.00 15 8.33 8.66 3.33 86.67 10.00 16.67 17.32 6.67 73.33 20.00 25.00 25.98 10.00 60.00 30.00 33.33 34.64 13.33 46.67 40.00 41.67 43.30 16.67 33.33 50.00 50.00 51.96 20.00 20.00 60.00 58.33 60.62 23.33 6.67 70.00 16 7.50 8.66 2.50 87.50 10.00 15.00 17.32 5.00 75.00 20.00 22.50 25.98 7.50 62.50 30.00 30.00 34.64 10.00 50.00 40.00 37.50 43.30 12.50 37.50 50.00 45.00 51.96 15.00 25.00 60.00 52.50 60.62 17.50 12.50 70.00 60.00 69.28 20.00 0.00 80.00 17 6.76 8.66 1.76 88.24 10.00 13.53 17.32 3.53 76.47 20.00 20.29 25.98 5.29 64.71 30.00 27.06 34.64 7.06 52.94 40.00 33.82 43.30 8.82 41.18 50.00 40.59 51.96 10.59 29.41 60.00 47.35 60.62 12.35 17.65 70.00 54.12 69.28 14.12 5.88 80.00 18 6.11 8.66 1.11 88.89 10.00 12.22 17.32 2.22 77.78 20.00 18.33 25.98 3.33 66.67 30.00 24.44 34.64 4.44 55.56 40.00 30.56 43.30 5.56 44.44 50.00 36.67 51.96 6.67 33.33 60.00 42.78 60.62 7.78 22.22 70.00 48.89 69.28 8.89 11.11 80.00 55.00 77.94 10.00 0.00 90.00 19 5.53 8.66 0.53 89.47 10.00 11.05 17.32 1.05 78.95 20.00 16.58 25.98 1.58 68.42 30.00 22.11 34.64 2.11 57.89 40.00 27.63 43.30 2.63 47.37 50.00 33.16 51.96 3.16 36.84 60.00 38.68 60.62 3.68 26.32 70.00 44.21 69.28 4.21 15.79 80.00 49.74 77.94 4.74 5.26 90.00 <<>>

Created a prototype SVM and neural net model for ternary phase diagrams through changing temperature in order to extrapolate and interpolate the phase behaviour of systems in temperature. <<>>

Weather Modelling:

Models and paper successfully submited to the publisher Alberta Academic Review.

August 5 — August 9

Weekly Summary

Clay Filtration:

Standard dilution curves were built for comparative extraction of chlorophyll. Results were taken at absorbances 649, 664, 670

<<>>

Explain T1, t2, t3, then explain what langmuir and freundlich are, not in detail but enough to know what is happening T1 fits langmuir and freundlich well - T2 (60) and T3 (100) don't - might be due to change in physical characteristic of pore site - equivalence in in binding energy is no longer valid. There is a model that described sigmoid-like adsorption, assumptions are different. Refered to as langmuir s-type or sigmoid type.

<<>>

Performed a second round of experiments on T1 with 0.5%, 1%, 2%, 2.5% and 5% loadings at T1. Isotherm results are given below. Still need an effective way to describe oil loss mathematically as a function. It seems that the chlorophyll absorbance data fits the langmur isotherm at this temp well. Still need to include the quality of fitting parameters in the scripts.

<<>>

Experimental results also have begun to show the relationship between %removal and temperature. Higher temperature, as predicted, allows improved removal of chlorophyll, but also causes a lower loss in oil. Thus, higher temperature operaiton has a higher oil recovery ratio. However it is more expensive to operate at a higher temperature - will attempt to adress this with economic trade off. A possible approach would be co-optimization of clay and protein bleaching. Still need to formulate continuum models. Possible approach would be to build solid/liquid separation into a black box model, and allow it to operate continously. Perform heat exchanger calculations off of slurry output.

<<>> <<>>

Phase Diagrams:

SVM hyperparameter (gamma and cost) optimization algorithms implemented in addition to the K optimizations for K nearest neighbours. Improved results seen. <<>>

Protein Modelling:

Weather Modelling:

Economic Modelling:

Started working on Stonks, the NN to predict canola futures. This is mostly an educational exercise as I wait for the lab to give data back on efficiency of the proteins. I'm using a MultiLayer Perceptron MLP to start off the project, then an Recursive Neural Network RNN. Later I plan to work integrate PCA to identify what causes the most dramatic shifts in the futures.

Also started looking into utility costs. Utilities can extrapolate the total cost of the operation. utilities generally are 15% of the cost of operation. using [Towler,_Gavin;_Sinnott,_Ray_K.]_Chemical_Engineer(z-lib.org).pdf

<<>>

May 6 — May 10

Weekly Summary

The 2019 Calgary iGEM team kicked off the summer by reviewing the medal requirements for this year's iGEM competition. We made sure all of our organizational tools (Mendeley, Benchling, etc.) were in place for the summer, and began chipping away at the mandatory laboratory safety courses which are required to work in University of Calgary labs. We also began literature review of a few potential project topics. The dry lab explored the feasability of a bioreactor CAD tool by reading through literature and talking with Dr. Dave Anderson, a bioinformatics professor at the University of Calgary. We also proposed developing a paper discovery engine that would suggest papers with contrasting ideas. This was received well by our team, so we decided to explore it further in following weeks. Monday , 05-06 Notes - Continued Literature review among the two possible projects, a bioreactor CAD tool, and a contradictory paper finder -Team meeting with the entire group. -Contacted multiple individuals in industry for HP listed below with their role. - Dale Schuurmans NLP proffessor @uAlberta James Wright NLP and Game Theory at Ualberta Robert Holte Machine Learning @uAlberta AltaML Machine Learning Company Richard Sutton Research Scientist @DeepMind Nick Savidov Rural Bioreactor Guru Aquatera Landfill Bioreactor users Enerkem Company that heavily uses bioreactors Ed Dziengielewski Landfill Manager - Peace Regional Landfill Deloitte Consulting Company Russ Greiner NLP, ML, Bioinformatics Proff @uAlberta Amii Too AI Research Group Darwin AI AI Company Derrick Rancourt Director @ Center for Genome Engineering Milo Koretsky Virtual Bioreactor Controller Marc Strous Leader of the Energy Bioengineering Group (Works with bioreactors) Dr Moshirpour AI help at UofC Reg Ouellette NLP help at successful start-up Adam Meyers Linguistics/CS prof at NYU, created Termolator meyers@cs.nyu.edu - -Developed ideas and relationship with Dr. Nick Savidov at lethbridge college towards bioreactor tool -looked into NLP and looked into algorithms - Worked heavily on the modelling portion of our project so far 4 models have been selected for further inquiry - Modelling of Degredation of Chlorophyll into its derivatives - Thermodynamic modelling of chlorophyl a and b -modelling proteins using SWISS server -Using a genetic algorithm for binding protein selection -Put in Work on Opentron application -theorized bioreactor module for opentron machine -Bioreactor idea, abandon the CAD (cannot compete with AutoCAD) and cannot compete with companies that sell them -Bioreactors are not inaccessible to farmers as they are very ingenious, but they don't have access to the high-tech versions of them through CAD or pre-built as they need cheap bulk, unlike Pharmaceutical companies that need quality. -Replace by just creating a spreadsheet to help farmers create their own bioreactors to be nutrient efficient. Modeling integrated into the spreadsheet/app/website. -wood chip bioreactors very expensive and not useful -Paper Discovery, Highly useful in the world of science as science is progressing faster than a person can keep up. Projects like CRAFT have been successful to summarize and figure out the current paradigm of say oncology. -Our project can certainly pursue comparisons/Contradictions between research papers. -Ethical Issue: Concentrates science into a dogma and new contradicting ideas that go against dogma will be discarded. - Toured the Life Sciences Innovation Hub -Met with stats prof (Andrew) BioReact - Johnson Su bioreactors seem to be the most popular composting for farmers -Meeting with Anderson (modeling) -Thoughts of splitting team into NLP and automation -Experimented with NLTK -Explored different lemmatizers -Discussed how one defines "constrast" - ultimately decided that we can't look for direct contrast, we should instead look at distance in similarity -Update iGEM Calgary to include last years information and show the current team -Worked on method to convert research papers to .txt based documents - PDF to txt done, text just needs a word-splitter -Project Summaries -Abandoned Bioreactors in meeting. -Lots of learning done -NLTK by software team - Updated igemCalgary.ca to reflect current team and added 2018 project summary Bioreactor idea highly invested into cumulating into meeting friday to decide on the project - Bioreactors as a main project abandoned. Started learning NLTK for paperTrail

Team:Calgary/Journal

Notebook / ENTRIES

+ Press for Options

Notebook

Entries

May 6 — May 10

Weekly Summary

May 13 — May 17

Weekly Summary

May 20 — May 24

Weekly Summary

May 27 — May 31

Weekly Summary

June 3 - June 7

Weekly Summary

June 10 - June 14

Weekly Summary

June 15 - June 19

Weekly Summary

June 24 - June 28

Weekly Summary

July 2 - July 5

Weekly Summary

July 8 - July 12

Weekly Summary

July 15 - July 19

Weekly Summary

July 22 - July 26

Weekly Summary

July 22 - July 26

Weekly Summary

July 29 - July 31

Weekly Summary

July 29 - July 31

Weekly Summary

August 5 - August 9

Weekly Summary

August 5 - August 9

Weekly Summary

August 19 - August 23

Weekly Summary

August 26 - August 31

Weekly Summary

September 3 — September 6

Weekly Summary

September 9 — September 15

Weekly Summary

September 16 — September 22

Weekly Summary

September 23 — September 29

Weekly Summary

October 1 — October 6

Weekly Summary

October 7 — October 13

Weekly Summary

May 20 — May 24

Weekly Summary

May 27 — May 31

Weekly Summary

June 3 — June 7

Weekly Summary

June 10 — June 14

Weekly Summary

June 17 — June 21

Weekly Summary

June 24 — June 28

Weekly Summary

July 1 — July 5

Weekly Summary

July 8 — July 12

Weekly Summary

July 15 — July 19

Weekly Summary

July 22 — July 28

Weekly Summary

July 29 — August 3

Weekly Summary

August 5 — August 10

Weekly Summary

August 12 — August 18