BME100 s2015:Group13 12pmL6

From OpenWetWare
Jump to navigationJump to search
BME 100 Spring 2015 Home
Lab Write-Up 1 | Lab Write-Up 2 | Lab Write-Up 3
Lab Write-Up 4 | Lab Write-Up 5 | Lab Write-Up 6
Course Logistics For Instructors
Wiki Editing Help


Name: Riley Baranek
Name: Trisha Dasgupta
Name: Tina Monteilh
Name: Payson Wallach
Name: Robel Okbe


Bayesian Statistics

Overview of the Original Diagnosis System 34 teams, each consisting of 6 students, diagnosed a total of 68 patients using the Open PCR machine and a fluorimeter test. Each team was given 2 patients to analyze. In order to prevent error each team was given 3 replicate sample tubes containing DNA for each patient. In order to improve accuracy a negative control, which did not contain the disease marker, SNP, and a positive control, containing the SNP were used for comparison. Before the actual patient samples and controls were tested using the fluorimeter, it was calibrated using the SYBR Green I solution and 6 DNA samples each with different concentrations of calf thymus DNA. 3 samples were used for each DNA concentration, so there were a total of 18 drop images analyzed for the calibration. The images of each of the calibration samples were then analyzed using ImageJ, and then the data collected from this was used to create a calibration curve which plotted calf thymus DNA concentration vs. the specific integrated density values. After this was completed the 3 replicate sample tubes for Patient 1, the 3 replicate sample tubes for Patient 2, and the controls were also tested using the fluorimeter. These drop images, 24 in total, were then analyzed in the same way using Image J, and by using the area, mean pixel value, and raw integrated density values, the concentrations of each of the samples were able to be calculated by plugging the values into the equation for the line of the calibration curve. These values for the patients were then compared to the concentration values of the negative and positive controls in order to predict if the patient was positive or negative. The positive control and negative control concentration values, concentration values for all 3 replicates for each patient, and the final conclusions were then added into a general spread sheet for each of the 34 teams. There were 20 patient's data that were not able to be used in data analysis since they were considered unreliable. Out of these 20, 8 were blank data in the spreadsheet, and the other 12 were inconclusive results. This left the other 32 patient's results to be used, for a total of 32 successful conclusions.

What Bayes Statistics Imply about This Diagnostic Approach

The Bayes value for calculation 1 was very close to 1.00, so this showed that it was reliable in its sensitivity in detecting the disease SNP. The Bayes value for the second calculation was also very
close to 1, so this showed that it was reliable in its specificity for detecting the disease. A possible source of human error could have been contamination of the samples, and this could
have happened if a pipet tip was not properly replaced. Error could have also happened if the tubes were mixed up or incorrectly labelled. A machine/device error could have caused
the results to be inaccurate if the PCR machine wasn't set up properly or if the machine itself was working incorrectly.

The Bayes value for calculation 3 was very small and so was the Bayes value for the 4th calculation. This shows the the PCR was not reliable in predicting the disease.

Computer-Aided Design

TinkerCAD is an internet app based 3D modeling software used to create basic designs for objects. The PCR machine used in the lab is open source, so all the pieces of the design are already created in CAD software and were available for free to add to TinkerCAD. However, our design required many large modifications, so the OpenPCR model was only used as a reference, and not actually modified.

Our Design

This design is significantly different from the OpenPCR design. In our design, the main modification is the addition of an internal operating system. This PCR no longer must be connected to a computer to operate,and this simplifies use and decreases set up time. This is controlled by the number pad on the front of the machine, which would also have presets for basic PCR tests, along with an input for custom tests. The other modification is the door on the machine. Instead of a hinged flap on the top that must be screwed down, a door is much simpler for inserting test samples. Both of these modifications were chosen based on the groups experience with the OpenPCR machine. The PCR machine had trouble connecting to the computer and would not send the test parameters via USB cable. We had to use another machine to fix this error. An internal system for programming the machine would reduce this chance of error, and overall simplify use. The door design was chosen because the current design does not show exactly if the samples are sealed in correctly, and a door would remove this ambiguity.

Feature 1: Consumables Kit

Consumables: In the original design, it was very inconvenient to have to wrap the sample tubes which contained SYBR green solution in foil. The SYBR green solution cannot be exposed to light as accurate data is desired, however it is not a completely manageable setup to just wrap those sample tubes in foil. That being said, we would like to introduce sample tubes that are not clear and block out light. If the SYBR green solution were to be packaged in these, the likelihood of it being exposed to light is decreased and the chance for error is significantly reduced.

Feature 2: Hardware - PCR Machine & Fluorimeter

Open PCR Machine and Software: In the original design, it was very difficult to run the tests as well as difficult to close the lid. The act of screwing the lid was a challenge, making the machine less accessible. That being said, we would like to introduce a hinged opening to the machine, making it easy to open and close. Easier insertion and removal of the sample tubes would be ideal. The fact that the old machine is made of wood is an issue due to the high internal temperatures as well, which the new design addresses and fixes. In addition, we would like to add a screen on the machine. Overall, these changes would make the machine more user friendly.

The Fluorimeter System: In the original design, the system lacked a mount for the camera and the results could have been quite inaccurate due to the slight exposures to light while opening and closing the lid. In addition, taking all the images from the camera and analyzing them via Image-J was definitely inconvenient as well as time consuming. That being said, we would like to introduce a built-in mount for a camera that could hold any smartphone. A hinged lid would be ideal as well, so light cannot enter the system as easily. And in general, it would be best to make the system much smaller. We would like to add a photo detector into the system, which will send the images of the solutions to another device via Bluetooth, making the system much more efficient.