rabadanlab / randomly Goto Github PK

A Library for Denoising Single-Cell Data with Random Matrix Theory

License: MIT License

Makefile 0.03% Jupyter Notebook 99.02% Python 0.95%

single-cell data-analysis random-matrix-theory sparsity denoising dropout single-cell-omics single-cell-rna-seq single-cell-analysis

randomly's People

Contributors

Stargazers

Watchers

Forkers

luisaparicio mbordyuh kraji vikarojkova huning2009 geoheil kp-hironaka grosenicklab hoangmgh

randomly's Issues

What is the reason for the abs() in this line of code?

In the refining computation, you project the genes into the random eigenspaces in order to identify those whose components differ from a normal distribution by a lot. Why do you take the absolute value of the eigenvector components?

randomly/randomly/randomly.py

Line 808 in 7961402

genesproj = np.dot(self.X3.T, (abs(self.Vr))[:, :]).T

Why do you use L_mp in part of this function, and L in another?

Why does line 311 reference L_mp (the eigenvalues that are consistent with the fitted MP distribution) and line 312 references L (all the eigenvalues). Shouldn't these be the same?

randomly/randomly/randomly.py

Lines 309 to 319 in 7961402

 def _tw(self): 

 '''Tracy-Widom critical eignevalue''' 

 gamma = self._mp_parameters(self.L_mp)['gamma'] 

 p = len(self.L) / gamma 

 sigma = 1 / np.power(p, 2/3) * np.power(gamma, 5/6) * \ 

 np.power((1 + np.sqrt(gamma)), 4/3) 

 lambda_c = np.mean(self.L_mp) * (1 + np.sqrt(gamma)) ** 2 + sigma 

 self.gamma = gamma 

 self.p = p 

 self.sigma = sigma 

 return lambda_c

Loading sparse matrix or anndata object

Hi,

Is loading a sparse matrix or anndata object functionality in the works?

https://anndata.readthedocs.io/en/stable/

Thanks!

Exporting corrected count matrix instead of scaled data

Randomly version: 0.1.0
Python version: 3.6.3
Operating System: Linux x86_64

Description

The output from Randomly has mean of ~ 0 for each gene (preprocess step centered data around zero after log2(1+TPM) transformation and filter). I am wondering if it is possible to add additional function to export a matrix that mimics a count matrix (some of the downstream packages I am using prefer to use count matrix as input.)

Thank you!

rabadanlab / randomly Goto Github PK

randomly's People

Contributors

Stargazers

Watchers

Forkers

randomly's Issues

What is the reason for the abs() in this line of code?

Why do you use L_mp in part of this function, and L in another?

Loading sparse matrix or anndata object

Exporting corrected count matrix instead of scaled data

Description

Recommend Projects

React

Vue.js

Typescript

TensorFlow

Django

Laravel

D3

Recommend Topics

javascript

web

server

Machine learning

Visualization

Game

Recommend Org

Facebook

Microsoft

Google

Alibaba

D3

Tencent

	def _tw(self):
	'''Tracy-Widom critical eignevalue'''
	gamma = self._mp_parameters(self.L_mp)['gamma']
	p = len(self.L) / gamma
	sigma = 1 / np.power(p, 2/3) * np.power(gamma, 5/6) * \
	np.power((1 + np.sqrt(gamma)), 4/3)
	lambda_c = np.mean(self.L_mp) * (1 + np.sqrt(gamma)) ** 2 + sigma
	self.gamma = gamma
	self.p = p
	self.sigma = sigma
	return lambda_c