BME100 s2018:Group11 W0800 L6

From OpenWetWare
Jump to: navigation, search
Owwnotebook icon.png BME 100 Spring 2018 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
BME494 Asu logo.png


Name: Kathryn Douglass
Name: Emily Glagolev
Name: Kayla Zeien
Name: Jonathon Chasteen


G11 8am 2017logo.png


Bayesian Statistics

Overview of the Original Diagnosis System
11 teams of approximately 4 students each diagnosed 22 patients total. Each team diagnosed 2 patients using the same process. First, a PCR reaction was created for each patient: PCR Reaction mix was added to each patient's DNA sample and put into a thermal cycler. Next, the dye SBYR Green 1 and a simple smart phone fluorimeter set-up was used to give teams a visual signal as to the presence of disease-associated SNP DNA. The images taken by the smartphone were quantitatively analyzed using image J, and then used to calculate the PCR product (i.e disease-associated SNP DNA) concentration. In order to do this accurately, the fluorimeter was manually calibrated using different concentrations of calf-thymus DNA. Each team also observed and analyzed a positive and negative control using the fluorimeter set-up, and the values found for each patient were compared to the positive and negative thresholds in order to determine the patient's diagnosis. In an attempt to try and prevent error, three replicate DNA samples were prepared and analyzed for each patient. In addition, for each sample (calibration concentrations, positive control, negative control, patient 1, patient 2, etc.) three separate pictures were taken by the smartphone and then analyzed. By recreating results multiple times, the effects of human error were minimized.

The results from each team were reported and used to calculate Bayesian statistics. While 22 patients were tested, Team #9 was unable to submit results on time and thus patient #55307 and patient #75503 were disregarded during the final statistical analysis. Out of the 20 patients included, 19 were diagnosed (the PCR results of patient #35518 were deemed inconclusive).

Our group encountered few problems, however the fact that each team executed the experiment slightly differently may have influenced final results.

What Bayes Statistics Implies about This Diagnostic Approach

Overall the PCR test results aligned with the team's conclusion in the 90 percentile for both positive and negative results. Given how close to 100% both calculations 1 and 2 were, the PCR test proved to be reliable. Tests yielding negative results proved to be slightly more reliable than the positive mainly due to the sample size of the negative results. Out of the total sample size, 70 percent tested negative compared to only 25 percent of the sample size testing positive. Having a larger pool of samples would allow for more reliable test conclusions.

Conclusion three gave about a 75% chance that if you are positive you will get the disease. Conclusion four was more reliable giving about 85% chance to not get the disease if the result was negative. The confidence of the Bayesian Statistics is effected by the population tested and the results found can be effected by population used. Overall PCR testing proved to be a somewhat reliable way to diagnosing patients.

The PCR process descibed above was prone to various errors both due to human error and machine/device error. One such error may have resulted from the sensitivity and accuracy the smartphone camera used when taking images of the droplets. If the images taken did not successfully capture the concentration of green within each droplet then the PCR product concentration calculated would be inaccurate potentially leading to a false conclusion. Another source of error may have occured due to the the flimsy design of the box when taking the picture of the droplets. When placing the camera inside the box light would have shone on the droplet and even with the box closed there were still slivers of light that could enter the box. This light could have affected the concentration of green the camera captured from the droplet, caused more of a glare in the image or caused the light-sensitive Syber Green 1 to react in a negative manner. These errors would have affected the final PCR product concentration reading found with the program Image J. Several sources of human error may have occurred throughout the experiment. One probable error would have been failing to collect all of the PCR sample and Syber Green 1 from the tube when forming each droplet. With inconsistent droplet sizes between each sample the final PCR product concentrations would have been affected leading to misrepresentative comparisons between the samples and slightly inaccurate final measurements.

Intro to Computer-Aided Design

3D Modeling
Our team chose to use Solid Works, a computer-aided engineering and design software. Since members of the team had had previous experience using Solid Works, the design process went relatively smoothly. Screenshots of the final design can be found below. One issue encountered however, was the complexity of the adjustable platform height aspect of the design. While in reality this component would involve mechanical operations (a screw connected to a knob to raise and lower the platform for example), with the limited solid works knowledge of the team this component required simplification. In the end, the adjustable height was represented by merely a knob (i.e. the cylinder in the top left corner on the front face of the device). Over all, however, the Solid Works software provided an accurate portrayal of our device design.

Our Design

Front View (closed lid)
Side View (closed lid)
Top View (open lid)

Our device was designed to solve one systematic problem with the original fluorimeter apparatus. During the experiment a smartphone was used to capture images of the droplets. Since the droplets had to be in a dark environment, a light box was used to cover the fluorimeter and the camera, meaning that each time a picture was taken the camera had to have a timer set and the box had to be repositioned over the system before that timer went off. This process was extremely tedious--it had to be done three times for each droplet. Indeed, since the apparatus was repositioned before each image, it likely affected the accuracy of the final results. To resolve this hassle, our design implements a camera holder on the outside of the box. Capable of adjusting the height with a basic crank mechanism using a knob, a simple ledge positions the phone camera precisely to capture images at the right distance from the droplet. The box has an open hole just for the phone camera. A combination of the phone and a piece of fabric hanging over it (attached via velcro strip) cover the rest of the hole, blocking light from entering the box. In addition, the box is designed with a rubber bottom to keep it from slipping when the phone is adjusted. Unlike the original fluorimeter, the top of our design is removable instead of the front. Made to attach magnetically, a removable top allows the drops to be easily added to the glass plate without having to move/adjust the entire apparatus. With this new design the whole system is connected. With the exception of the actual PCR droplets, no aspect of the design requires repositioning. Since the phone is placed in a specified slot, is no need to measure and remeasure the distance between the phone and the slide. In other words, our design improves upon the original by keeping the experimental environment fixed, removing the variability between trials.

Feature 1: Consumables

Our consumables kit will include slides that will fit in our fluorimeter (size 2.5cm x 7.5cm), SYBR green solution in our newly designed darkened tubes, buffer, PCR mix , and plastic tubes with labels and the desired amount of primer in them.

Major weaknesses we encountered were the labeling system of the tubes and using foil to keep the SYBR Green dark. To combat these issues we designed label stickers that can be put on the tubes. These will come with positive and negative stickers as well as stickers with sample and trials numbers. It will also include colors to help further organize the tubes. This is shown in the picture below. We found the foil covering the SYBR green tubes was not an efficient way to do so. To fix this issue we also designed a dark tube to not allow light to get to the SYBR green.

PCR stickers11.jpg

Feature 2: Hardware - PCR Machine & Fluorimeter

Although the Flourimeter system and the PCR machines functioned well enough to provide reliable results, there were flaws that could be improved to make the process more efficient. Adjusting the height of the fluorimeter apparatus was the most inefficient aspect of the process. Stacking plastic boxes to adjust for height created an unstable platform as well as caused less accuracy in flushing the camera to the droplets. The cell phone stand also created issues. Long skinny phones such as the Samsung Galaxy 8 proved difficult to remain stable. The stand would also slide into to different positions after handling the phone to send and take pictures, requiring constant readjustment to ensure a consistent distance between the phone and the droplets. The system also did a poor job at keeping the fluorimeter concealed from light.The box's lid that covered the fluorimeter did not close all the way, sometimes affecting the quality of the images. The Open PCR machine's weaknesses were tied to its long heating process. The machines requirement of large PCR tubes and inefficient heat containing lid assisted in its unnecessary lengthy process.

To address these issues, the Fluorimeter system will incorporate a more efficient smartphone holder where the smartphone will remain in a fixed position until completion of the experiment, this will eliminate any issued with realignment and refocusing of the camera. The phone itself will be height adjustable to ensure the stability of the apparatus and to create a simplified way to flush the camera and droplet. With the phone in a fixed position on the face of the system, a removable magnetic lid will be on top. The removable magnetic top will create a tight seal, concealing the fluorimeter completely from light. To make a more time efficient PCR machine, a seal will line the bottom of the lid to better contain the heat during the PCR process. Also, the machine will allow a much smaller sized PCR tube to fit in the machine to increase the speed of the reaction.