HNSCPredict :- A Tool for Identification of HNSCC from Single Cell Genome using Deep Learning

A computational approach tool to predict Head and Neck Cancer affected patients from their single cell RNA seq data.

Introduction

Head and neck cancer, which encompasses a range of malignancies affecting the respiratory tract and upper digestive tract and also the seventh most common cancer in the world.

This tool aims to use Artifical Neural Network (Deep Learning) model to classify Normal Control(NC) patients and Head and Neck Cancer (HNSCC) patients from their single cell RNA seq data. The tool takes 10x single cell genomics data as input and predicts whether the patient is diseased or healthy with the help of highly trained model.

An excellent feature selection method called mRMR (Minimum Redundancy Maximum Relevance) was used to find out top 100 features which act as promising biomarkers in classification and prediction of Normal and Diseased patients.

Please read/cite the content about the AlzScPredict for complete information including algorithm behind the approach.

Installation

Install my-project with pip

  pip install HNSCPred

You if previously installed please update the python package to the latest version using the command below

  pip3 install --upgrade HNSCPred

Usage/Examples

After installation of the HNSCPred package in your python enviornment. Import the library using the below code.

import HNSCPred

The HNSCPred comes with 1 inbuilt modules .

• Predict Please import all 3 modules in your python enviornment using the code below.

from HNSCPred import preprocessing
from HNSCPred import Model_Selection
from HNSCPred import PredictionModule

After importing all the important pre requisites. You can follow the demo below for your case.

import pandas as pd
df = pd.read_csv("Your file path here")

#Takes single cell data in form of dataframe with cells in rows and features in columns. Returns preprocessed dataframe.
processed = preprocessing.preprocess_data(df)

# Load the model with the code below. This takes no arguments.
my_model = Model_Selection.load_model()

# Prediction:- Execute the code below to get the output. It takes 2 arguments i.e 
# preloaded model and the processed dataframe with 35 features.

PredictionModule.predict_patient(my_model,processed)

Used By

This project is used by the following companies:

• Company 1
• Company 2

Authors

• Aman Srivastava.
• Akanksha Jarwal.
• Prof. G.P.S. Raghava

Acknowledgements

• Awesome Readme Templates
• Awesome README
• How to write a Good readme