liwencong1995 / SDS-Capstone-Zebrafish

SDS 410: SDS Capstone

Zebrafish

Introduction

This repository was created to store the work that was done by Shuli Hu, Dejia Tang, Wencong (Priscilla) Li, and JiYoung (Justine) Yun in SDS 410: SDS Capstone Class in Spring 2018.

This study focuses on the classification of biological creatures’ phenotypes. In this project, we utilize support vector machine to distinguish structures of Zebrafish’s brains by using data generated from landmark analysis (cited Morgan’s paper). We create a tool for biologists to intuitively classify three-dimensional biological shapes into two groups, usually defined as wild type and mutant, and understand which part of the shapes have the most impact on the classification result. This project derives from Professor Barresi’s biological image analysis research at Smith College.

Final Paper and Model

The PDF version of the final paper can be found in 8. Final Paper folder. In addition, the source code of both the Model and the User Interface can be downloaded from 9. Final Model folder.

3. Input Data

3. Input Data folder contains all datafiles used in the study.

• raw folder

  • AT_landmarks_3-28-18.csv and ZRF_landmarks_3-29-18.csv are the original data files of the two channels received from Morgan Schwartz.
  • AT_landmarks.csv and ZRF_landmarks.csv contains the renamed duplicated sample index 112 as 112_1 and 112_2. (SCRIPT: svm.ipynb)

• tidy folder

  • tidyLandmarks_AT.csv and tidyLandmarks_ZRF.csv are the data that have been cleaned and rearranged into a long format. (SCRIPT: tidyLandmark.R)
  • In landmark_AT_w_index.csv and landmark_ZRF_w_index.csv, landmark_index column was added so that we can use this new variable to run SVM models. (SCRIPT: addIndex.ipynb and visualization.R)

• median folder

  • In AT_median.csv and ZRF_median.csv, median and 2
  • median values were computed for replacing the NA values. (SCRIPT: median_calculation.R)
  • landmark_AT_median.csv and landmark_ZRF_median.csv

• final folder

  • landmark_AT_filled_w_median.csv and landmark_ZRF_filled_w_median.csv are the datasets that have the NA values replaced with the median r values of each landmark by data type (wildtype or mutant). (SCRIPT: svm.ipynb)
  • landmark_AT_filled_w_2median.csv and landmark_ZRF_filled_w_2median.csv are the datasets that have the NA values replaced with the 2
  • median r values of each landmark by data type (wildtype or mutant). (SCRIPT: svm.ipynb)