olenive Goto Github PK

accelerometer

compare_coverage_depth

A Julia script for comparing the coverage between two chromosomes for a set of sequenced DNA samples.

datahub

A centralized location for storing curated data ready for inclusion in cBioPortal.

deephyperneat

A public python implementation of the DeepHyperNEAT system for evolving neural networks. Developed by Felix Sosa and Kenneth Stanley. See paper here: https://eplex.cs.ucf.edu/papers/sosa_ugrad_report18.pdf

jsub

Tool for systematically creating and running jobs via job files on the LSF queuing system.

petritype

pixelgen

pytorch-neat

rapier.rs

Official website for the Rapier physics engine.

recommendation-tensorflow-model

two-state_model_benchmarking

Code used for benchmarking of the two-state model simulated using Gillespie or the Poisson sampler (Psampler) siumlation method described in Lenive et al 2016 (Inferring extrinsic noise from single-cell gene expression data using approximate Bayesian computation). Please note that this is intended as a benchmark only and is not a stand-alone tool in it's current version. An example file containing model parameters is found in the data directory. The path to this file is currently hard-coded. To run the benchmark: make ./benchmark.exe Depends on a C++ compiler and libgsl for random number generation (see Makefile). The model is hard coded in the Gillespie algorithm file and (unavoidably) in the Psampler. Aims: - Read nominal parameter vectors from rows in a text file. - Implement a noise function (specific to the 2-state model), the same as was used in inference procedure described in the manuscript. - Run simulations corresponding to a fixed number of samples of perturbed parameters obtained from the nominal parameters using the noise function. - Optional output to files of perturbed parameter vectors and final states and simulation times. - A benchmarking script that runs the above simulations. - Run benchmark with output to files and compare to see if Gillespie and Psampler produce similar results. - Run benchmark with minimal output to files so as not to conflate simulation time and I/O time.