"
Team:OUC-China/Part III
From 2013.igem.org
Error Analysis
Error Detection And Estimation
1.Detection of The System Errors in Bare Board
In order to get rid of the system errors from the bare board using to help us in the fluorescence measuring, we use simple regression to analyse the scanning bare board fluorescence value to find if there exists the system errors.
Table 1.The Bare Board’s Row Data
Figure 1. The Bare Board’s Residual Plot
The point which is not in the 95% confidence bounds shows in red ,which means abonormal value,and the ones in green means the point in the 95% confidence bounds.
Figure2.Raw data and Simple Regression Straight Line.
We plot 2 Simple Regression Straight Lines,one in red is the regression straight line of the original data,and another one in blue is the regression straight line after eliminating abnormal data.
Table2.The coefficient in two lines
The p1 and p3 is approximate to 0,so we can make sure our bare board will not take the system errors based on the the linear change which is in our experiment,
After that,we can belive only random errors in our experimental data not system errors.
2.Simulate The Noise In Our RNA Guardian
We use Gillespie algorithm to simulate the experiment ,and we find the Stochastic Oscillator is within the reasonable errors,but it may influence our observation value .We use the parameters we get in the model 1 to run the algorthm.And we get the result of the Simulated Control Group and the simulated Experimental Group3.
Figure 3.Simulated Control Group 
Figure 4.Simulated Experimental Group 3
Comparing with the Figure 3 and Figure 4,we can see the figure at the top right corner about the time-mRNA’s Number and the top left corner about the mRNA’s Number.In Simulated Control Group,the most term frequency corresponding to the number of mRNA is around 10,and in Simulated Experimental Group 3,that is about 16 relative increasing 60% than the Simulated Control Group,which agrees with our prediction in Model II.That is because we use the parameter - degradation rate of mRNA in Control Group we estimates is 0.5 relatively,and in Experimental Group is about 0.3,that is what we have get before.
And let’s look at the left bottom ,we can see the most term frequency corresponding to the number of protein in simulated Control Group is from 180 to 240,and in Simulated Experimental Group 3 it is around 370,relative increasing 54% comparing with the Simulated Control Group.
Next,we want to know if our experimental data exceeds the random errors we have simulated.We calculated the average value of the number of mRNA at each time point M in Control Group .The number of mRNA at each time point we called it R(i),i=1 to 1000.We define
Which means the number of mRNA at each time point ‘s turbulence,after that we calculate the average value of every B(i),it is about 0.1443.
Figure 5.The turbulence of mRNA at each time point
Table3. The Turbulence of The Data Measured in The Control Group
We calculated the turbulence of the data measured in the control group,and find NO.1 and NO.2 exceed 0.1443,that means the 2 groups may include not only random errors.
Table 4.Comparison of Conrol Group and Experimental Group 3(RBS0 + CDS + RBS1)
Let’s look at the Control Group and the Experimantal Group 3,we find all groups exceeds 0.1443,especially NO.1 and NO.6 are obvious,which means NO.1 and NO.6 exceed normal range as expected.
We compare the Control Group and the Experimental Group 2 (RBS1 + RBS0 + CDs) below:
We find only NO.2 is as expected,its relative increase exceed the stochastic noise as expected.
So,we get the conclusion considring all factors:
Comparing with Control Group ,The Stability of mRNA is sorted as:
(RBS0 + CDS + RBS1)(+) > (RBS1 + RBS0 + CDs)(+) > (RBS1 + RBS1 + RBS0 + CDs + RBS1)(-)
(RBS0 + CDS + RBS1)(+):60% relatively increase theoretically and in practice
(RBS1 + RBS0 + CDs)(+):20% relatively increase theoretically and in practice
(RBS1 + RBS1 + RBS0 + CDs + RBS1)(-) :About 80% lower according to raw data.
We have to make a couples of repeated experiments, some parameters in our model still not accuracy, which is necessay to be amended in our model .
Reference:
1. Experiment Design and Data Processing .2008.Southeast University Press.
2. A General Method for Numerically Simulating the Stochastic Time Evolution of Coupled Chemical Reactions. DANIEL T. GILLESPIE. JOURNAL OF COMPUTATIONAL PHYSICS 2, 403-434 (1976).
