Team:Austin UTexas/Results



We transformed 497 BioBricks from iGEM distribution kits into our burden monitor strain. We performed over 49 burden assays and analyzed the measurements through a pipeline of scripts to yield the following results. In the end, we obtained 392 distinct observations of the relevant measurements for quantifying burden for 330 different parts.

Our data is meant to provide evidence for the following:

  • BioBricks have burden associated with them.
  • There can be more than one type of burden imposed by a BioBrick.

We define burden as percent reduction in growth rate.

The peak mean specific growth rate and GFP rate for each experiment within the following data is normalized to a value of 1 for both variables. This multivariate normalization was conducted under the assumption that most BioBricks are not significantly burdensome. Therefore, statistically significant burdensome parts can be reported as such with surety.

BioBricks have burden associated with them.

Figure 1 shows shows that growth rate reduction can be associated with different constructs. It displays the descending order of normalized mean specific growth rate for each strain isolate that was tested. The progressively darker blue color correlates to the specific growth rate for each strain isolate shown. The more burdensome a construct is, the more reduced the growth rate will be for a host cell harboring the BioBrick. This directly provides evidence that BioBricks have a burden associated with them.

Figure 1. Graph of normalized specific growth rates for each strain isolate.

Again referencing our frequently aforementioned definition of burden as a percent reduction in growth rate, the burden value is simply the decimal value representing this percentage. The burden value is equivalent to a construct’s normalized growth rate subtracted from 1, with 1 being defined as the mean value of “no burden” constructs for the total distribution. For example, a part with a normalized growth rate of 0.75 would have a burden value of 0.25 and therefore contributes to a 25% reduction in growth rate.

Figure 2 is a density plot that illustrates the burden values (percent reduction in growth rate) that our tested constructs exhibited, under the assumption that most BioBricks aren’t burdensome. A construct with a negative or zero burden value means that the construct displayed a growth rate reduction of 0% and is therefore not burdensome. However, if a construct has a measured burden value of 0.4, then that indicates a 40% reduction in growth rate. Although excluded from the data, the grand means of our control strains are plotted to serve as a reference point from which one may assess the burden of other constructs. While most parts do show some minute burden value, most parts are not burdensome because they do not significantly contribute to a reduction in growth rate relative to the expected growth rate for that construct.

Figure 2. Density plot of the burden distribution of BioBricks.

There can be more than one type of burden imposed by a BioBrick.

To differentiate sources of burden, we also measured the GFP expression rate concurrently with growth rate in our “burden assays”. The linear regression is built from the measurements of our control strains. Our control strains provide a reasonable model to construct a threshold to describe translational burden (caused by ribosomal reallocation) because each of the five control strains have a different promoter and ribosome binding site combination. Therefore, we would expect to see a linear curve where stronger promoters with weaker ribosome binding sites exhibit a lower growth rate and a lower GFP expression rate as resources are allocated towards its plasmid, starving the genome and simulating an outcome of translational stress. As all combinations are used, we observed a linear regression model with a slope that is not significantly different from 1 as the threshold at which translational burden can be distinguished from other types of burden not associated with the mere stress of a construct’s translational elements.

The positive linear slope shows that constructs with a higher growth rate and concurrently higher GFP rate are less burdensome than those with a lower growth rate and lower GFP rate, and that 100% of the growth rate reduction associated with those parts is sourced from translational burden. Simply put, points lower on the regression line are more burdensome than points higher on the regression line, and the burden imposed on those parts is a result of translational stress.

However, if a construct shows a higher GFP rate than what would be linearly coordinated with its measured growth rate, the growth rate reduction reported must be attributed to something in addition to translational stress. In other words, a BioBrick of this type would have a growth rate less than what it should be as dictated by the linear model. Thus, in addition to translational burden, such a BioBrick also has another type of burden which further reduces its growth rate.
Figure 3 provides evidence that there can be more than one type of burden imposed by a construct. Significantly burdensome parts are distinguished from parts showing no significant burden and from parts exhibiting additional, “other” burden.

Figure 3. Scatter plot of BioBricks' GFP expression rates as a function of growth rates.


These results are important because they can help to better characterize constructs in the iGEM registry such that reliable parts may be distinguished from unreliable, burdensome ones. These results, when used in context with our model, may help a person decide which parts are better suited for their purposes, whether it be simply cloning a part in the lab or propagating the part for long-term, large-volume industrial use.

Further studies from these data could include investigation of: (1), patterns of burden, (2), if burdensome constructs bear genetic similarities, and (3), whether those faults may be improved upon to yield more stable constructs. Another worthwhile investigation would be a study in the loci of mutation for burdensome parts, indicating which areas of the construct are more unstable to bolster the former suggested study.

Limitations of our approach include the premise of our project – our control strains are purposefully engineered to be burdensome and are no exception to the rule that more burdensome parts are more unstable. Using our control strains long-term in this project presented the difficulty of maintaining a consistent measure, but even throughout the course of a year of using the same control strain stocks, the change we saw in the expected and measured burden for those parts was predictable.