biolearns 0.0.14

BioLearns: Computational Biology and Bioinformatics Toolbox in Python

biolearns

BioLearns: Computational Biology and Bioinformatics Toolbox in Python http://biolearns.com

Installation

• From PyPI

pip install biolearns

Documentation and Tutorials

• We select three examples listed below. For full list of tutorial, check our github wiki page:

  Wiki

1. Read TCGA Data

Example: Read TCGA Breast invasive carcinoma (BRCA) data

Data is downloaded directly from https://gdac.broadinstitute.org/. The results here are in whole or part based upon data generated by the TCGA Research Network: https://www.cancer.gov/tcga.

from biolearns.dataset.TCGA import TCGACancer

brca = TCGACancer('BRCA')
mRNAseq = brca.mRNAseq
clinical = brca.clinical

TCGA cancer table shortcut:

	Barcode	Cancer full name	Version
1	ACC	Adrenocortical carcinoma	2016_01_28
2	BLCA	Bladder urothelial carcinoma	2016_01_28
3	BRCA	Breast invasive carcinoma	2016_01_28
4	CESC	Cervical and endocervical cancers	2016_01_28
5	CHOL	Cholangiocarcinoma	2016_01_28
6	COAD	Colon adenocarcinoma	2016_01_28
7	COADREAD	Colorectal adenocarcinoma	2016_01_28
8	DLBC	Lymphoid Neoplasm Diffuse Large B-cell Lymphoma	2016_01_28
9	ESCA	Esophageal carcinoma	2016_01_28
...	...	...	...

2. Gene Co-expression Analysis

We firstly download and access the mRNAseq data.

from biolearns.dataset.TCGA import TCGACancer

brca = TCGACancer('BRCA')
mRNAseq = brca.mRNAseq

mRNAseq data is noisy. We filter out 50% of genes with lowest mean values, and then filter out 50% remained genes with lowest variance values.

from biolearns.preprocessing.filter import expression_filter
mRNAseq = expression_filter(mRNAseq, meanq = 0.5, varq = 0.5)

We then use lmQCM class to create an lmQCM object lobj.

The gene co-expression analysis is performed by simply call the fit() function.

from biolearns.coexpression.lmQCM import lmQCM

lobj = lmQCM(mRNAseq)
clusters, genes, eigengene_mat = lobj.fit()

3. Univariate survival analysis

We firstly download and access the mRNAseq data. Use breast cancer as an example.

from biolearns.dataset.TCGA import TCGACancer

brca = TCGACancer('BRCA')
mRNAseq = brca.mRNAseq

We import logranktest from survival subpackage. Choose gene "ABLIM3" as the univariate input.

from biolearns.survival import logranktest

r = mRNAseq.loc['ABLIM3',].values

We find the intersection of univariate, time, and event data

bcd_m = [b[:12] for b in mRNAseq.columns]
bcd_p = [b[:12] for b in clinical.index]
bcd = np.intersect1d(bcd_m, bcd_p)

r = r[np.nonzero(np.in1d(bcd, bcd_m))[0]]
t = brca.overall_survival_time[np.nonzero(np.in1d(bcd, bcd_p))[0]]
e = brca.overall_survival_event[np.nonzero(np.in1d(bcd, bcd_p))[0]]

We perform log-rank test:

logrank_results, fig = logranktest(r[~np.isnan(t)], t[~np.isnan(t)], e[~np.isnan(t)])
test_statistic, p_value = logrank_results.test_statistic, logrank_results.p_value

The output figure looks like: