KASE analysis demonstration

kase

Kinome Array Significance Evaluator

A tool to analyze Kinome Array data for scientific publication. Examples:

V1.0

Re-coded application from EXCEL macro, reducing computation from hours to seconds Runs 100,000 iterations of a simulation in ~10s

For Python 2.7 We recommend running using Anaconda for easy package install https://www.continuum.io/downloads

PLANNED FUTURE FEATURES:

Figure generation
Configurable random seed

Purpose

Kinome Arrays often give confusing results because each detector corresponds to multiple kinases. Therefore, using simulation, this program approximates a parametric analysis of the distribution of number of hits for each kinase given that x substrates are chosen, where x equals the actual number of substrates chosen. Using this approximated distribution, a z-score and subsequent p-value (two-tailed) is calculated for each kinase.

An inexact approximation: Standard normal distribution is used as simplifying assumption; although we expect a skewed distribution due ot proximity to zero. While a large n=100,000 (random) simulations can be accomodated, this is a tiny fraction of all the true combinations. 141 C 10 ~= 617 trillion Nonetheless, this is sufficent to rank and compare effect size.

Instructions

Run kase_gui.py for interactive analysis Note that three files are included as example input/output (in order to load in program) SubstrateKinaseDefinition.csv SigSubstrates.csv output.tsv