←

Overview

OD Measurements

DoE

Media Optimisation

Model Fit

Interaction Profiles

Model Results

Parameters

For science to be dependable and reproducible, good measurement is essential,so good measurements are essential for our iGEM project with Yarrowia lipolytica.First, we checked optical density (OD) for both transformed and untransformed strains, which helped us understand their growth dynamics. Then, we worked on media optimization, based on the principles of Design of Experiments (DoE) to find the best conditions for good growth and performance.. We also dug deeper, checking the fatty acid profile with LCMS and examining the hydrocarbon profile in our transformed strain using GCMS. These measurements were very insightful while moving forward with our project.

OD600 measurements, also known as optical density at 600 nanometers, refer to a common method used in microbiology and molecular biology to quantify the concentration of microorganisms, such as bacteria or yeast, in a liquid culture. The measurement is based on the principle that microbial cells, when suspended in a liquid medium, can scatter and absorb light.

We wanted to check whether there was any difference between the growth rate of our transformed and untransformed strain as we needed information about the OD pattern for other experments like western blot. So we decided to measure the growth rate using OD600 measurement. We measured the OD of both the strains simultaneously for 34 hours straight. We aim to track the growth over longer time periods to better understand the growth pattern.

We found that the transformed strain was growing at a faster pace than our untransformed strain. The transformed Po1g strain freached its stationary phase of 14.8 within less than 30 hours. While our untransformed strain was still only at half the OD of 7.2 even after 34 hours. We couldnt find a reason why this was happening, this needs to be studied with further replicative experiments.

Seeing the results we wanted to know whether we could increase the growth rate and thereby achieve a larger cell density in a shorter time span thereby increasing the chances of achieving stationary phase earlier than normal, this would help us since our hydrocarbons are only produced in the stationary phase; helping us extract the more hydrocarbons in a shorter span also. Also we found from literature that higher C/N ratio leads to more lipid So, we decided to do a media optimization with DoE principles to find the most conducive environment for the robust growth and proliferation of our yeast cells. This was done by tweaking the C/N ratio by changing the glucose and peptone content in the media.

According to the American Society for Quality, DoE is defined as a branch of applied statistics that deals with planning, conducting, analyzing, and interpreting controlled tests to evaluate the factors that control the value of a parameter or group of parameters. It allows for multiple input factors to be manipulated, determining their effect on a desired output (response). By manipulating multiple inputs at the same time, DOE can identify important interactions that may be missed when experimenting with one factor at a time (OFAT). We used the JMP platform for the design of experiments and its analysis.

We decided to use the Response Surface Methodology (RSM) to understand the interaction between our explanatory variables and inturn how it effects our response variable. We choose the amount of the three constituents of our yeast growth media ie. Glucose, peptone, yeast extract as our explanatory variables and RPM was considered as the 4 explanatory variable or factor. The OD600 after 24H was chosen as the response variable.

We tweaked the C/N ratio by changing the glucose and peptone content in the media as a higher C/N leads to more lipid production. The OD600 after 24H was chosen as the response variable and RPM,glucose,peptone and yeast extract were chosen as the factors to create a RSM plot . JMP was used for generating the custom design needed for the experiments. We wanted to do a definitive screening first to determine all the factors affecting the growth but couldn’t perform so due to time and resources constraints.

The RPM was varied from 150 - 250 RPM, glucose was varied from 0.3 - 0.7g, peptone from 0.3 - 0.6g and yeast extract from 0.2 - 0.5g in a total of 10mL volume. Then a custom DoE design was generated in JMP and the following table developed.

We used the ‘custom design’ DoE tool in JMP and the model parameters are in the image below:

As seen in the image above for model parameters, we took into account the indivdual effect of all the factors and also the cross interaction between the different factors followed by quadratic interactions. We couldnt perform a full factorial design due to time and resource constraints.

The following experiments in the table were then performed in the wet lab and the OD600 measurements after 24h was then loaded on the table.

After fitting the model with least squares fit, we removed two quadartic interactions (glucose*glucose and peptone*peptone) as they have a p-value above 0.7 which is very insignificant.

We achieve an R^2 = 0.98644 for the model which is a pretty good fit with a p-value of 0.0178 which is pretty significant too. This shows that our model is able to explain the results we got in wet lab in a very good way.

From the image above it is pretty evident that all the interactions are significant as they have a p-value greater than 0.05. But for simplicity purposes I am only concentrating on the effects which have a p-value less than 0.01. The rest of the effects can also be explained in similar ways.

The interaction profile results can help understand which all factor are interacting with each other and effecting the response variables. This helps us to then knockout insignificant interactions and factors from our model helping in effective screening. If two lines are intersecting or tending to intersect , then they are said to be interacting with each other. In the above image, we can see that almost all factors we choose are important factors and interacting with each other.

RSM or Response Surface Methodology, is a statistical and mathematical technique used in experimental design and optimization. RSM involves creating plots, known as RSM plots, to visualize the relationship between multiple input variables and a response or output variable. These plots help identify optimal conditions and understand how changes in input factors impact the desired outcome. By systematically exploring the response surface, RSM enables efficient problem-solving and process improvement.

The RSM was plotted using the above code to obtain the prediction:

According to the effects summary, peptone concentration has the most significance in p-value measures. Visualizing the Peptone concentration effect with the surface plot showed:

The surface plot implies that peptone follows a linear relationship, that is, increasing the peptone amount lead to a higher OD600 atleast in the range from 0.3-0.7g per 10mL.

According to the effects summary, glucose alone has a p-value of 0.032, which is still significant since it is below 0.05. Visualizing the result by surface plot shows that:

Glucose also follows a linear relationship, leading to a higher OD with increase in concentration of glucose atleast in the range of 0.3-0.7g per 10mL. This may change when higher amount of glucose are taken.

According to the effects summary, it has a p value of 0.029, after visualizing by surface plot we see:

The yeast extract follows the following pattern where lower concentration of yeast extract gave the maximum OD but on increasing the concentration further leading to a lower OD only to start increasing again after approaching a concentration of 0.5g in 10mL.

RPM has a P value of 0.027 which is pretty significant. Visualizing the interaction via surface plot showed:

The result showed that with increasing RPM the OD keeps on increasing and reaching a maximum around 200-220 RPM and then starts decreasing again. Showing that aeration alone beyond a point is not just enough to increase the cell density.

This interaction gave a very significant p-value of 0.0086, and is the most significant interaction. Visualizing by surface plot and contour plot.

The red bands shows higher OD values and blue bands the lower values of OD. The surface and contour plot shows the following interactions and the effect on OD (In increasing OD values starting from the least to the highest):

- High glucose-Low Peptone
- Low glucose-Low peptone
- Low glucose-High peptone
- High glucose-High peptone

This shows that peptone may be the limiting factor in this case of glucose vs peptone.

The RPM-Yeast extract cross interaction gave a p-value of 0.0092, which is very significant. Visualizing the interaction via surface and contour plot:

The red bands shows higher OD values and blue bands the lower values of OD. The highest OD was seen at high RPM and low yeast extract. The lowest OD was seen at high RPM and medium concentration(0.3g-0.4g) of yeast extract. We cannot conclude which one would be the limiting factor in this case as we can infer from the surface plot.

The effect summary gave a p-value of 0.011 to the cross interaction between glucose and yeast extract. Visualizing using surface and contour plots:

According to the analysis, low yeast extract and low glucose gave the highest OD value, followed by high glucose and high yeast extract.

Over here we have done only the analysis of the most significant cross factor interactions.There are other interactions also which can also be visualized similarly using surface and contour plots.

Using the prediction profiler on JMP, we maximize the desirability to get the parameters needed for the maximum OD with the present factor levels.

The prediction profiler gave the following as the optimal parameters for a 10mL culture:

- RPM - 243
- Glucose - 0.7g
- Peptone - 0.6g
- Yeast Extract - 0.2g

If the above parameters are implemented we are expected to get an OD600 around 34.45 in 24 hours within a confidence interval of 27.88-41.031.