Skip to content

Commit 9600bd3

Browse files
pfhpfh
committed
Minor cleanup.
1 parent 613be7e commit 9600bd3

File tree

3 files changed

+9
-7
lines changed

3 files changed

+9
-7
lines changed

‎DESCRIPTION‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -10,7 +10,7 @@ Description: Transform RNA-Seq count data so that variance due to biological
1010
Authors@R: person("Paul", "Harrison", email = "[email protected]", role = c("aut", "cre"))
1111
Maintainer: Paul Harrison <[email protected]>
1212
URL: https://github.com/MonashBioinformaticsPlatform/varistran
13-
Version: 1.0.0
13+
Version: 1.0.1
1414
License: LGPL-2.1 | file LICENSE
1515
Depends:
1616
grid

‎README.md‎

Lines changed: 6 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -2,12 +2,12 @@
22

33
Varistran is an R package providing a Variance Stabilizing Transformation appropriate for RNA-Seq data, and a variety of diagnostic plots based on such transformation.
44

5+
* [Function reference](http://logarithmic.net/varistran/reference/index.html)
6+
57
* [Online demo](http://rnasystems.erc.monash.edu:3838/pfh/2015/demo-varistran)
68

79
* [A slideshow describing Varistran](http://rnasystems.erc.monash.edu:3838/pfh/2016/varistran/)
810

9-
* [Function reference](http://logarithmic.net/varistran/reference/index.html)
10-
1111
* [Poster for ABACBS 2015](doc/varistran-poster-abacbs-2015.pdf) [(on F1000, doi: 10.7490/f1000research.1110757.1)](http://f1000research.com/posters/4-1041)
1212

1313
Varistran is developed by Paul Harrison ([email protected], [@paulfharrisson](https://twitter.com/paulfharrison)) for the [Monash Bioinformatics platform](https://platforms.monash.edu/bioinformatics/).
@@ -44,7 +44,7 @@ y <- varistran::vst(counts, design=design)
4444

4545
By default, Anscombe's variance stabilizing transformation for the negative binomial distribution is used. This behaves like log2 for large counts (log2 Counts-Per-Million if `cpm=T` is given).
4646

47-
An appropraite dispersion is estimated with the aid of the design matrix. (If omitted, this defaults to a column of ones, for blind estimation of the dispersion. This might slightly over-estimate the dispersion. A third possibility is to estimate the dispersion with edgeR.)
47+
An appropraite dispersion is estimated with the aid of the design matrix. If omitted, this defaults to a column of ones, for blind estimation of the dispersion. This might slightly over-estimate the dispersion. A third possibility is to estimate the dispersion with edgeR.
4848

4949
### Diagnostic plots
5050

@@ -84,7 +84,7 @@ varistran::shiny_report(counts=counts)
8484
* [Online demo](http://rnasystems.erc.monash.edu:3838/pfh/2015/demo-varistran)
8585

8686

87-
### Tests
87+
## Test suite
8888

8989
After downloading the source code, a suite of tests can be run with:
9090

@@ -100,3 +100,5 @@ Outputs are places in a directory called `test_output`.
100100
* [Monash Bioinformatics Platform, Monash University](https://platforms.monash.edu/bioinformatics)
101101

102102
* [RNA Systems Laboratory, Monash University](http://rnasystems.erc.monash.edu)
103+
104+

‎paper.md‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -1,5 +1,5 @@
11
---
2-
title: "Varistran: Anscombe's variance stabilizing transformation for RNA-Seq gene expression data"
2+
title: "Varistran: Anscombe's variance stabilizing transformation for RNA-seq gene expression data"
33
tags:
44
- RNA-Seq
55
- gene expression
@@ -18,7 +18,7 @@ bibliography: paper.bib
1818

1919
# Summary
2020

21-
RNA-Seq measures RNA expression levels in a biological sample using high-throughput cDNA sequencing, producing counts of the number of reads aligning to each gene. Noise in RNA-Seq read count data is commonly modelled as following a negative binomial distribution, where the variance is a quadratic function of the expression level. However many statistical, machine learning, and visualization methods work best when the noise in a data set has equal variance. Varistran is an R package that uses Anscombe's [-@Anscombe1948] variance stabilizing transformation for the negative binomial distribution to transform RNA-Seq count data, so that the noise has equal variance across all measured gene expression levels. The transformed data may be treated as log~2~ transformed gene expression levels, but with variability reduced at low read counts. Varistran also includes a function to open a Shiny report with simple diagnostic visualizations, including a plot to assess how effective the variance stabilization was, a biplot of samples and genes, and a heatmap. This allows defective samples, sample mislabling, and batch effects to be easily identified.
21+
RNA-seq measures RNA expression levels in a biological sample using high-throughput cDNA sequencing, producing counts of the number of reads aligning to each gene. Noise in RNA-seq read count data is commonly modelled as following a negative binomial distribution, where the variance is a quadratic function of the expression level. However many statistical, machine learning, and visualization methods work best when the noise in a data set has equal variance. Varistran is an R package that uses Anscombe's [-@Anscombe1948] variance stabilizing transformation for the negative binomial distribution to transform RNA-seq count data, so that the noise has equal variance across all measured gene expression levels. The transformed data may be treated as log~2~ transformed gene expression levels, but with variability reduced at low read counts. Varistran also includes a function to open a Shiny report with simple diagnostic visualizations, including a plot to assess how effective the variance stabilization was, a biplot of samples and genes, and a heatmap. This allows defective samples, sample mislabling, and batch effects to be easily identified.
2222

2323
# References
2424

0 commit comments

Comments
 (0)