Normal donot vs CVID using CyTOF Number as count per ml generated using flowJo Predictions were performed using the caret library
This is the data generated using CyTOF, performed on NK cell panel. The study has 26 normal donor (ND) and 16 common variable immunodeficiency (CVID) samples. The data was first generated by traditional analysis, by making biaxial plot of each possible population, using flowjo dongle. Once the populations were identified, the number per uL were genrated for each population identied (columns) versus each sampl (row). The csv file was imported into the R and dataframe was created. Data was split between 80% training set and 20% test set. Using caret and supporting libraries various models were generated. After checking acccuracy of each model the prediction was performed on test data. This is a basic program intended to show serve as proof of concept that the final product from our study can further be utlized for developing and predictive model, which ultimately be used for predicting the disease and its stages. In this analysis no null imputation was required. Also the scaling was not required owing to the fact that most numbers fell within the comparable range. Finally, only few models have been tested, each representative of coomon categories. However in theory multiple models can be tested as long as it is supported by the framework.