Our Rochester team has five modeling projects, two software projects, and four hardware projects for iGEM 2020. To learn more about our project, see our wiki: https://2020.igem.org/Team:Rochester.
To improve endometriosis diagnostics, our team created a model to assess endometriosis risk from clinical variables by using machine learning on a dataset of 756 patients. The "Cleaning Predictive Model Data.R" script was used to transform the raw dataset into a form readable by the machine learning script "Endometriosis Predictive Model.R".
Our team developed lateral flow assays (LFA) to measure levels of endometriosis biomarkers in menstrual blood, thereby diagnosing endometriosis. The Sensitivity & Specificity, Antibody, and LFA Model were created in support of assay development.
There are 12 biomarkers for endometriosis in peripheral blood and menstrual effluent reported in literature. Our modeling team used combined log odds ratios to find the best combination of three biomarkers that contribute the most to the diagnostic accuracy of our test panel.
An LFA requires two antibodies that can bind to the target biomarker simultaneously. We used Rosetta software to predict the epitopes of 14 candidate antibodies for four biomarkers, and identified antibody pairs to be used in our assay. The workflow is "relax.sh", "prepack.sh", "snugdock.sh".
In designing our LFAs, we developed a model to find the optimal test line position and reagent concentrations, using the "Rochester_LFA_Model.m" script. We planned to parameter fit our experimental data to our model using the "Rochester_LFA_Model_DataFit.m" script.
To develop endometriosis diagnostics, our team designed an estrogen sensing circuit. We created an ODE model of the circuit to find the optimal promoter strength and plasmid copy number for this circuit.
To support future iGEM teams in the Diagnostics Track, we created a software collecting biomarkers that have been used in iGEM. Users can contribute to the database or search for potential biomarkers of the disease they want to study.
To promote endometriosis diagnosis, we integrated our Clinical Predictive Model into a web UI where users can fill out an anonymous survey, and find out their risk of endometriosis from their answers.
In support of LFA development, we built a smartphone-based platform to image and quantify the signal of our LFA. We used Arduino to control an LED in this platform using a potentiometer.
In an effort to make our diagnostics accessible to clinics without a lab, we built a DIY centrifuge to prepare sample for our LFA. We used Arduino to control the centrifuge's speed using a keypad with the connection to the ESC controller and LCD screen.