Medullo Classifier Shiny App

Authors:	Komal Rathi
Contact:	rathik@email.chop.edu
Organization:	D3B, CHOP
Status:	This is "work in progress"
Date:	2024-10-13

Introduction

R Shiny application built on the Medulloblastoma subtypes classifier R package: https://github.com/d3b-center/medullo-classifier-package

The app can be accessed here: https://komalrathi.shinyapps.io/medulloclassifiershinyapp

Input files

The application requires two inputs:

1. Expression Matrix with HUGO symbols as rows and Sample identifiers as columns.
   • A dataframe of NxM dimension containing expression values.
   • Rownames are HUGO/HGNC gene symbols and column names are Sample identifiers.
   • Allowed input types: (1) FPKM (2) TPM (3) Quantile normalized data (4) microarray data (RMA).
   • Allowed formats: Tab delimited text, RData and RDS files.

	Sample_1	Sample_2	Sample_3	Sample_n
HUGO_Gene_1
HUGO_Gene_2
HUGO_Gene_3

2. Metadata with subtype information.
   • A vector of length M containing Medulloblastoma subtypes corresponding to each sample identifier.
   • Tab delimited format should contain one subtype per line and no headers.
   • RData and RDS formats should be character vector of subtypes.
   • Allowed values are Group3, Group4, SHH, WNT and U (for Unknown)
   • Allowed formats: Tab delimited text, RData and RDS files.

NOTE: Both expression and meta files should have the same extension type

Output

Output consists of four tables:

1. Predicted subtypes and P-values for each sample.
2. Confusion Matrix
3. Overall Accuracy statistics
4. Class based statistics

For more details, please refer to: https://github.com/d3b-center/medullo-classifier-package