# GSVA / ssGSEA command-line interface and Python module
The GSVA (gene-set variance analysis) package from R bioconductor provides efficient computation of single-sample gene-set enrichment analysis (ssGSEA). This pakcage provides a python implmented CLI, and Python module with Pandas inputs and outputs, as well as a docker to run this R package.
* Repository is here: https://github.com/jason-weirather/GSVA
* Autodoc manual is here: https://jason-weirather.github.io/GSVA/
##### Disclaimer
I am not the creator or author of GSVA. This is a CLI and python hook created to make their package easy to use from the command line and python. *This is not the offical site for the GSVA bioconductor package*
Find the official R package here
https://doi.org/doi:10.18129/B9.bioc.GSVA
##### And if you find this useful, please cite the author's publication
Hänzelmann S, Castelo R and Guinney J (2013). “GSVA: gene set variation analysis for microarray and RNA-Seq data.” BMC Bioinformatics, 14, pp. 7. doi: 10.1186/1471-2105-14-7, http://www.biomedcentral.com/1471-2105/14/7.
## Quickstart - CLI through docker
### Execute GSVA in docker
Just be careful to let docker know all the volumes you need to mount. Here we will do everything in our current directory.
1. Start with an expression csv with gene-wise rows and sample-wise columns
```
$ cat example_expression.csv | cut -f 1-3 -d ',' | head -n 6
gene_name,S-1,S-2
MT-CO1,13.852,12.328
MT-CO2,13.405999999999999,12.383
MT-CO3,13.234000000000002,12.109000000000002
MT-ATP8,13.805,11.789000000000001
MT-ATP6,13.5,11.703
```
2. Use any gene sets in **gmt** format where each row follows the convention `name <tab> description <tab> gene1 <tab> gene2 <tab> ... <tab> geneN`
```
cat c2.cp.v6.0.symbols.gmt | head -n 6 | cut -f 1-5
KEGG_GLYCOLYSIS_GLUCONEOGENESIS http://www.broadinstitute.org/gsea/msigdb/cards/KEGG_GLYCOLYSIS_GLUCONEOGENESIS ACSS2 GCK PGK2
KEGG_CITRATE_CYCLE_TCA_CYCLE http://www.broadinstitute.org/gsea/msigdb/cards/KEGG_CITRATE_CYCLE_TCA_CYCLE IDH3B DLST PCK2
KEGG_PENTOSE_PHOSPHATE_PATHWAY http://www.broadinstitute.org/gsea/msigdb/cards/KEGG_PENTOSE_PHOSPHATE_PATHWAY RPE RPIA PGM2
KEGG_PENTOSE_AND_GLUCURONATE_INTERCONVERSIONS http://www.broadinstitute.org/gsea/msigdb/cards/KEGG_PENTOSE_AND_GLUCURONATE_INTERCONVERSIONS UGT1A10 UGT1A8 RPE
KEGG_FRUCTOSE_AND_MANNOSE_METABOLISM http://www.broadinstitute.org/gsea/msigdb/cards/KEGG_FRUCTOSE_AND_MANNOSE_METABOLISM MPI PMM2 PMM1
KEGG_GALACTOSE_METABOLISM http://www.broadinstitute.org/gsea/msigdb/cards/KEGG_GALACTOSE_METABOLISM GCK GALK1 GLB1
```
3. Run GSVA
```
$ docker run -v $(pwd):$(pwd) vacation/gsva:1.0.4 \
GSVA --gmt $(pwd)/c2.cp.v6.0.symbols.gmt \
$(pwd)/example_expression.csv \
--output $(pwd)/example_pathways.csv
```
##### You're done. Thats it. Enjoy, check your output.
Running outside of docker on your system is just as easy (actually easier) but you need to have the required programs installed (see below).
```
$ cat example_pathways.csv | cut -f 1-3 -d ',' | head -n 6
name,S-1,S-2
BIOCARTA_41BB_PATHWAY,0.0686308398590944,0.257169127694153
BIOCARTA_ACE2_PATHWAY,0.11082238459933501,-0.22231034473486602
BIOCARTA_ACH_PATHWAY,0.514192767265737,0.149291024991685
BIOCARTA_ACTINY_PATHWAY,-0.0144944990305252,0.407870971955071
BIOCARTA_AGPCR_PATHWAY,0.6224821629523449,-0.0128449355033173
```
### For more advanced options you can list the options
```
$ docker run vacation/gsva:1.0.4 GSVA -h
usage: GSVA [-h] [--tsv_in] --gmt GMT [--tsv_out] [--output OUTPUT]
[--meta_output META_OUTPUT] [--method {gsva,ssgsea,zscore,plage}]
[--kcdf {Gaussian,Poisson,none}] [--abs_ranking] [--min_sz MIN_SZ]
[--max_sz MAX_SZ] [--parallel_sz PARALLEL_SZ]
[--parallel_type PARALLEL_TYPE] [--mx_diff MX_DIFF] [--tau TAU]
[--ssgsea_norm SSGSEA_NORM] [--verbose]
[--tempdir TEMPDIR | --specific_tempdir SPECIFIC_TEMPDIR]
input
Execute R bioconductors GSVA
optional arguments:
-h, --help show this help message and exit
Input options:
input Use - for STDIN
--tsv_in Exepct CSV by default, this overrides to tab (default:
False)
--gmt GMT GMT file with pathways (default: None)
Output options:
--tsv_out Override the default CSV and output TSV (default:
False)
--output OUTPUT, -o OUTPUT
Specifiy path to write transformed data (default:
None)
--meta_output META_OUTPUT
Speciify path to output additional run information
(default: None)
command options:
--method {gsva,ssgsea,zscore,plage}
Method to employ in the estimation of gene-set
enrichment scores per sample. By default this is set
to gsva (Hanzelmann et al, 2013) and other options 6
gsva are ssgsea (Barbie et al, 2009), zscore (Lee et
al, 2008) or plage (Tomfohr et al, 2005). The latter
two standardize first expression profiles into
z-scores over the samples and, in the case of zscore,
it combines them together as their sum divided by the
square-root of the size of the gene set, while in the
case of plage they are used to calculate the singular
value decomposition (SVD) over the genes in the gene
set and use the coefficients of the first right-
singular vector as pathway activity profile. (default:
gsva)
--kcdf {Gaussian,Poisson,none}
Character string denoting the kernel to use during the
non-parametric estimation of the cumulative
distribution function of expression levels across
samples when method="gsva". By default,
kcdf="Gaussian" which is suitable when input
expression values are continuous, such as microarray
fluorescent units in logarithmic scale, RNA-seq log-
CPMs, log-RPKMs or log-TPMs. When input expression
values are integer counts, such as those derived from
RNA-seq experiments, then this argument should be set
to kcdf="Poisson". This argument supersedes arguments
rnaseq and kernel, which are deprecated and will be
removed in the next release. (default: Gaussian)
--abs_ranking Flag used only when mx_diff=TRUE. When
abs_ranking=FALSE (default) a modified Kuiper
statistic is used to calculate enrichment scores,
taking the magnitude difference between the largest
positive and negative random walk deviations. When
abs.ranking=TRUE the original Kuiper statistic that
sums the largest positive and negative random walk
deviations, is used. In this latter case, gene sets
with genes enriched on either extreme (high or low)
will be regarded as'highly' activated. (default:
False)
--min_sz MIN_SZ Minimum size of the resulting gene sets. (default: 1)
--max_sz MAX_SZ Maximum size of the resulting gene sets. (default:
None)
--parallel_sz PARALLEL_SZ
Number of processors to use when doing the
calculations in parallel. This requires to previously
load either the parallel or the snow library. If
parallel is loaded and this argument is left with its
default value (parallel_sz=0) then it will use all
available core processors unless we set this argument
with a smaller number. If snow is loaded then we must
set this argument to a positive integer number that
specifies the number of processors to employ in the
parallel calculation. (default: 0)
--parallel_type PARALLEL_TYPE
Type of cluster architecture when using snow.
(default: SOCK)
--mx_diff MX_DIFF Offers two approaches to calculate the enrichment
statistic (ES) from the KS random walk statistic.
mx_diff=FALSE: ES is calculated as the maximum
distance of the random walk from 0. mx_diff=TRUE
(default): ES is calculated as the magnitude
difference between the largest positive and negative
random walk deviations. (default: True)
--tau TAU Exponent defining the weight of the tail in the random
walk performed by both the gsva (Hanzelmann et al.,
2013) and the ssgsea (Barbie et al., 2009) methods. By
default, this tau=1 when method="gsva" and tau=0.25
when method="ssgsea" just as specified by Barbie et
al. (2009) where this parameter is called alpha.
(default: None)
--ssgsea_norm SSGSEA_NORM
Logical, set to TRUE (default) with method="ssgsea"
runs the SSGSEA method from Barbie et al. (2009)
normalizing the scores by the absolute difference
between the minimum and the maximum, as described in
their paper. When ssgsea_norm=FALSE this last
normalization step is skipped. (default: True)
--verbose Gives information about each calculation step.
(default: False)
Temporary folder parameters:
--tempdir TEMPDIR The temporary directory is made and destroyed here.
(default: /tmp)
--specific_tempdir SPECIFIC_TEMPDIR
This temporary directory will be used, but will remain
after executing. (default: None)
```
## Installation
#### Method 1: Install on your system
1. Install R https://www.r-project.org/
2. Install the R bioconductor packaqge GSEABase and GSVA
```
$ Rscript -e 'source("http://bioconductor.org/biocLite.R");\
library(BiocInstaller);\
biocLite(pkgs=c("GSEABase","GSVA"),dep=TRUE)'
```
3. Install this package `$ pip install GSVA`
#### Method 2: Run GSVA via the docker
`$ docker pull vacation/gsva:latest`
## Use GSVA Python CLI in your python code
First install GSVA Python CLI on your system as described above. For details on the `gsva(expression_df,genesets_df,...)` function parameters see https://jason-weirather.github.io/GSVA/
### Workflow example - Go from an expression-based tSNE plot to a pathway-based tSNE plot in a Jupyter notebook
Here we will convert a per-sample per-gene expression matrix to a per-sample per-pathway enrichment matrix. We will plot the values using tSNE.
These code snipits and outputs are from a Jupyter notebook.
```python
import pandas as pd
from GSVA import gsva, gmt_to_dataframe
# Some extras to look at the high dimensional data
from plotnine import *
from sklearn.manifold import TSNE
```
Read in a Broad reference pathway gmt file. Notice the "member" and "name" fields. If you make your own dataframe to use, these are the required column names.
```python
genesets_df = gmt_to_dataframe('c2.cp.v6.0.symbols.gmt')
genesets_df.head()
```
| | description | member | name |
|---|---------------------------------------------------|--------|---------------------------------|
| 0 | http://www.broadinstitute.org/gsea/msigdb/card... | ACSS2 | KEGG_GLYCOLYSIS_GLUCONEOGENESIS |
| 1 | http://www.broadinstitute.org/gsea/msigdb/card... | GCK | KEGG_GLYCOLYSIS_GLUCONEOGENESIS |
| 2 | http://www.broadinstitute.org/gsea/msigdb/card... | PGK2 | KEGG_GLYCOLYSIS_GLUCONEOGENESIS |
| 3 | http://www.broadinstitute.org/gsea/msigdb/card... | PGK1 | KEGG_GLYCOLYSIS_GLUCONEOGENESIS |
| 4 | http://www.broadinstitute.org/gsea/msigdb/card... | PDHB | KEGG_GLYCOLYSIS_GLUCONEOGENESIS |
This example has 200 samples
```python
expression_df = pd.read_csv('example_expression.csv',index_col=0)
expression_df.iloc[0:5,0:5]
```
| gene_name | S-1 | S-2 | S-3 | S-4 | S-5 |
|-----------|--------|--------|--------|--------|--------|
| MT-CO1 | 13.852 | 12.328 | 13.055 | 11.898 | 10.234 |
| MT-CO2 | 13.406 | 12.383 | 13.281 | 11.578 | 11.156 |
| MT-CO3 | 13.234 | 12.109 | 13.352 | 11.531 | 10.422 |
| MT-ATP8 | 13.805 | 11.789 | 13.414 | 11.883 | 11.141 |
| MT-ATP6 | 13.500 | 11.703 | 13.227 | 11.219 | 10.836 |
```python
XV = TSNE(n_components=2).\
fit_transform(expression_df.T)
df = pd.DataFrame(XV).rename(columns={0:'x',1:'y'})
(ggplot(df,aes(x='x',y='y'))
+ geom_point(alpha=0.2)
)
```
The default command runs without verbose message output. but take notice, that genes that are not part of the `expression_df` are dropped from the analysis, and depending on your choice of GSVA method, genes for which there is not enough expression (i.e. all zero expression) will be dropped.
```python
pathways_df = gsva(expression_df,genesets_df)
pathways_df.iloc[0:5,0:5]
```
| name | S-1 | S-2 | S-3 | S-4 | S-5 |
|-------------------------|-----------|-----------|-----------|----------|-----------|
| BIOCARTA_41BB_PATHWAY | 0.068631 | 0.257169 | -0.146907 | 0.020151 | -0.234537 |
| BIOCARTA_ACE2_PATHWAY | 0.110822 | -0.222310 | -0.161572 | 0.370659 | -0.003318 |
| BIOCARTA_ACH_PATHWAY | 0.514193 | 0.149291 | 0.226279 | 0.289960 | 0.016071 |
| BIOCARTA_ACTINY_PATHWAY | -0.014494 | 0.407871 | -0.062163 | 0.055607 | 0.424726 |
| BIOCARTA_AGPCR_PATHWAY | 0.622482 | -0.012845 | 0.317349 | 0.286368 | 0.022540 |
```python
YV = TSNE(n_components=2).\
fit_transform(pathways_df.T)
pf = pd.DataFrame(YV).rename(columns={0:'x',1:'y'})
(ggplot(pf,aes(x='x',y='y'))
+ geom_point(alpha=0.2)
)
```
Raw data
{
"_id": null,
"home_page": "https://github.com/jason-weirather/GSVA",
"name": "GSVA",
"maintainer": "",
"docs_url": null,
"requires_python": "",
"maintainer_email": "",
"keywords": "bioinformatics,R,enrichment,GSVA,ssGSEA,GSEA,bioconductor",
"author": "Jason L Weirather",
"author_email": "jason.weirather@gmail.com",
"download_url": "https://files.pythonhosted.org/packages/e5/64/75cbd888bbd1574a16f89a49bdaf66d394c3d088b40929a334179d51770f/GSVA-1.0.6.tar.gz",
"platform": "",
"description": "# GSVA / ssGSEA command-line interface and Python module\n\nThe GSVA (gene-set variance analysis) package from R bioconductor provides efficient computation of single-sample gene-set enrichment analysis (ssGSEA). This pakcage provides a python implmented CLI, and Python module with Pandas inputs and outputs, as well as a docker to run this R package.\n\n* Repository is here: https://github.com/jason-weirather/GSVA\n* Autodoc manual is here: https://jason-weirather.github.io/GSVA/\n\n##### Disclaimer\n\nI am not the creator or author of GSVA. This is a CLI and python hook created to make their package easy to use from the command line and python. *This is not the offical site for the GSVA bioconductor package*\n\nFind the official R package here\n\nhttps://doi.org/doi:10.18129/B9.bioc.GSVA\n\n##### And if you find this useful, please cite the author's publication\n\nH\u00e4nzelmann S, Castelo R and Guinney J (2013). \u201cGSVA: gene set variation analysis for microarray and RNA-Seq data.\u201d BMC Bioinformatics, 14, pp. 7. doi: 10.1186/1471-2105-14-7, http://www.biomedcentral.com/1471-2105/14/7.\n\n## Quickstart - CLI through docker\n\n### Execute GSVA in docker\n\nJust be careful to let docker know all the volumes you need to mount. Here we will do everything in our current directory.\n\n1. Start with an expression csv with gene-wise rows and sample-wise columns\n\n```\n$ cat example_expression.csv | cut -f 1-3 -d ',' | head -n 6 \ngene_name,S-1,S-2\nMT-CO1,13.852,12.328\nMT-CO2,13.405999999999999,12.383\nMT-CO3,13.234000000000002,12.109000000000002\nMT-ATP8,13.805,11.789000000000001\nMT-ATP6,13.5,11.703\n```\n\n2. Use any gene sets in **gmt** format where each row follows the convention `name <tab> description <tab> gene1 <tab> gene2 <tab> ... <tab> geneN`\n\n```\ncat c2.cp.v6.0.symbols.gmt | head -n 6 | cut -f 1-5\nKEGG_GLYCOLYSIS_GLUCONEOGENESIS\thttp://www.broadinstitute.org/gsea/msigdb/cards/KEGG_GLYCOLYSIS_GLUCONEOGENESIS\tACSS2\tGCK\tPGK2\nKEGG_CITRATE_CYCLE_TCA_CYCLE\thttp://www.broadinstitute.org/gsea/msigdb/cards/KEGG_CITRATE_CYCLE_TCA_CYCLE\tIDH3B\tDLST\tPCK2\nKEGG_PENTOSE_PHOSPHATE_PATHWAY\thttp://www.broadinstitute.org/gsea/msigdb/cards/KEGG_PENTOSE_PHOSPHATE_PATHWAY\tRPE\tRPIA\tPGM2\nKEGG_PENTOSE_AND_GLUCURONATE_INTERCONVERSIONS\thttp://www.broadinstitute.org/gsea/msigdb/cards/KEGG_PENTOSE_AND_GLUCURONATE_INTERCONVERSIONS\tUGT1A10\tUGT1A8\tRPE\nKEGG_FRUCTOSE_AND_MANNOSE_METABOLISM\thttp://www.broadinstitute.org/gsea/msigdb/cards/KEGG_FRUCTOSE_AND_MANNOSE_METABOLISM\tMPI\tPMM2\tPMM1\nKEGG_GALACTOSE_METABOLISM\thttp://www.broadinstitute.org/gsea/msigdb/cards/KEGG_GALACTOSE_METABOLISM\tGCK\tGALK1\tGLB1\n```\n\n3. Run GSVA\n\n```\n$ docker run -v $(pwd):$(pwd) vacation/gsva:1.0.4 \\\n GSVA --gmt $(pwd)/c2.cp.v6.0.symbols.gmt \\\n $(pwd)/example_expression.csv \\\n --output $(pwd)/example_pathways.csv\n```\n\n##### You're done. Thats it. Enjoy, check your output.\n\nRunning outside of docker on your system is just as easy (actually easier) but you need to have the required programs installed (see below). \n\n```\n$ cat example_pathways.csv | cut -f 1-3 -d ',' | head -n 6\nname,S-1,S-2\nBIOCARTA_41BB_PATHWAY,0.0686308398590944,0.257169127694153\nBIOCARTA_ACE2_PATHWAY,0.11082238459933501,-0.22231034473486602\nBIOCARTA_ACH_PATHWAY,0.514192767265737,0.149291024991685\nBIOCARTA_ACTINY_PATHWAY,-0.0144944990305252,0.407870971955071\nBIOCARTA_AGPCR_PATHWAY,0.6224821629523449,-0.0128449355033173\n```\n\n\n### For more advanced options you can list the options\n\n```\n$ docker run vacation/gsva:1.0.4 GSVA -h\nusage: GSVA [-h] [--tsv_in] --gmt GMT [--tsv_out] [--output OUTPUT]\n [--meta_output META_OUTPUT] [--method {gsva,ssgsea,zscore,plage}]\n [--kcdf {Gaussian,Poisson,none}] [--abs_ranking] [--min_sz MIN_SZ]\n [--max_sz MAX_SZ] [--parallel_sz PARALLEL_SZ]\n [--parallel_type PARALLEL_TYPE] [--mx_diff MX_DIFF] [--tau TAU]\n [--ssgsea_norm SSGSEA_NORM] [--verbose]\n [--tempdir TEMPDIR | --specific_tempdir SPECIFIC_TEMPDIR]\n input\n\nExecute R bioconductors GSVA\n\noptional arguments:\n -h, --help show this help message and exit\n\nInput options:\n input Use - for STDIN\n --tsv_in Exepct CSV by default, this overrides to tab (default:\n False)\n --gmt GMT GMT file with pathways (default: None)\n\nOutput options:\n --tsv_out Override the default CSV and output TSV (default:\n False)\n --output OUTPUT, -o OUTPUT\n Specifiy path to write transformed data (default:\n None)\n --meta_output META_OUTPUT\n Speciify path to output additional run information\n (default: None)\n\ncommand options:\n --method {gsva,ssgsea,zscore,plage}\n Method to employ in the estimation of gene-set\n enrichment scores per sample. By default this is set\n to gsva (Hanzelmann et al, 2013) and other options 6\n gsva are ssgsea (Barbie et al, 2009), zscore (Lee et\n al, 2008) or plage (Tomfohr et al, 2005). The latter\n two standardize first expression profiles into\n z-scores over the samples and, in the case of zscore,\n it combines them together as their sum divided by the\n square-root of the size of the gene set, while in the\n case of plage they are used to calculate the singular\n value decomposition (SVD) over the genes in the gene\n set and use the coefficients of the first right-\n singular vector as pathway activity profile. (default:\n gsva)\n --kcdf {Gaussian,Poisson,none}\n Character string denoting the kernel to use during the\n non-parametric estimation of the cumulative\n distribution function of expression levels across\n samples when method=\"gsva\". By default,\n kcdf=\"Gaussian\" which is suitable when input\n expression values are continuous, such as microarray\n fluorescent units in logarithmic scale, RNA-seq log-\n CPMs, log-RPKMs or log-TPMs. When input expression\n values are integer counts, such as those derived from\n RNA-seq experiments, then this argument should be set\n to kcdf=\"Poisson\". This argument supersedes arguments\n rnaseq and kernel, which are deprecated and will be\n removed in the next release. (default: Gaussian)\n --abs_ranking Flag used only when mx_diff=TRUE. When\n abs_ranking=FALSE (default) a modified Kuiper\n statistic is used to calculate enrichment scores,\n taking the magnitude difference between the largest\n positive and negative random walk deviations. When\n abs.ranking=TRUE the original Kuiper statistic that\n sums the largest positive and negative random walk\n deviations, is used. In this latter case, gene sets\n with genes enriched on either extreme (high or low)\n will be regarded as'highly' activated. (default:\n False)\n --min_sz MIN_SZ Minimum size of the resulting gene sets. (default: 1)\n --max_sz MAX_SZ Maximum size of the resulting gene sets. (default:\n None)\n --parallel_sz PARALLEL_SZ\n Number of processors to use when doing the\n calculations in parallel. This requires to previously\n load either the parallel or the snow library. If\n parallel is loaded and this argument is left with its\n default value (parallel_sz=0) then it will use all\n available core processors unless we set this argument\n with a smaller number. If snow is loaded then we must\n set this argument to a positive integer number that\n specifies the number of processors to employ in the\n parallel calculation. (default: 0)\n --parallel_type PARALLEL_TYPE\n Type of cluster architecture when using snow.\n (default: SOCK)\n --mx_diff MX_DIFF Offers two approaches to calculate the enrichment\n statistic (ES) from the KS random walk statistic.\n mx_diff=FALSE: ES is calculated as the maximum\n distance of the random walk from 0. mx_diff=TRUE\n (default): ES is calculated as the magnitude\n difference between the largest positive and negative\n random walk deviations. (default: True)\n --tau TAU Exponent defining the weight of the tail in the random\n walk performed by both the gsva (Hanzelmann et al.,\n 2013) and the ssgsea (Barbie et al., 2009) methods. By\n default, this tau=1 when method=\"gsva\" and tau=0.25\n when method=\"ssgsea\" just as specified by Barbie et\n al. (2009) where this parameter is called alpha.\n (default: None)\n --ssgsea_norm SSGSEA_NORM\n Logical, set to TRUE (default) with method=\"ssgsea\"\n runs the SSGSEA method from Barbie et al. (2009)\n normalizing the scores by the absolute difference\n between the minimum and the maximum, as described in\n their paper. When ssgsea_norm=FALSE this last\n normalization step is skipped. (default: True)\n --verbose Gives information about each calculation step.\n (default: False)\n\nTemporary folder parameters:\n --tempdir TEMPDIR The temporary directory is made and destroyed here.\n (default: /tmp)\n --specific_tempdir SPECIFIC_TEMPDIR\n This temporary directory will be used, but will remain\n after executing. (default: None)\n```\n\n## Installation\n\n#### Method 1: Install on your system\n\n1. Install R https://www.r-project.org/ \n2. Install the R bioconductor packaqge GSEABase and GSVA \n\n```\n$ Rscript -e 'source(\"http://bioconductor.org/biocLite.R\");\\\n library(BiocInstaller);\\\n biocLite(pkgs=c(\"GSEABase\",\"GSVA\"),dep=TRUE)'\n```\n\n3. Install this package `$ pip install GSVA`\n\n#### Method 2: Run GSVA via the docker\n\n`$ docker pull vacation/gsva:latest`\n\n## Use GSVA Python CLI in your python code\n\nFirst install GSVA Python CLI on your system as described above. For details on the `gsva(expression_df,genesets_df,...)` function parameters see https://jason-weirather.github.io/GSVA/ \n\n### Workflow example - Go from an expression-based tSNE plot to a pathway-based tSNE plot in a Jupyter notebook\n\nHere we will convert a per-sample per-gene expression matrix to a per-sample per-pathway enrichment matrix. We will plot the values using tSNE.\n\nThese code snipits and outputs are from a Jupyter notebook.\n\n\n```python\nimport pandas as pd\nfrom GSVA import gsva, gmt_to_dataframe\n# Some extras to look at the high dimensional data\nfrom plotnine import *\nfrom sklearn.manifold import TSNE\n```\n\nRead in a Broad reference pathway gmt file. Notice the \"member\" and \"name\" fields. If you make your own dataframe to use, these are the required column names.\n\n```python\ngenesets_df = gmt_to_dataframe('c2.cp.v6.0.symbols.gmt')\ngenesets_df.head()\n```\n\n|\t| description\t | member | name |\n|---|---------------------------------------------------|--------|---------------------------------|\n| 0\t| http://www.broadinstitute.org/gsea/msigdb/card... | ACSS2 | KEGG_GLYCOLYSIS_GLUCONEOGENESIS |\n| 1\t| http://www.broadinstitute.org/gsea/msigdb/card... | GCK | KEGG_GLYCOLYSIS_GLUCONEOGENESIS |\n| 2\t| http://www.broadinstitute.org/gsea/msigdb/card... | PGK2 | KEGG_GLYCOLYSIS_GLUCONEOGENESIS |\n| 3\t| http://www.broadinstitute.org/gsea/msigdb/card... | PGK1 | KEGG_GLYCOLYSIS_GLUCONEOGENESIS |\n| 4\t| http://www.broadinstitute.org/gsea/msigdb/card... | PDHB | KEGG_GLYCOLYSIS_GLUCONEOGENESIS |\n\nThis example has 200 samples\n\n```python\nexpression_df = pd.read_csv('example_expression.csv',index_col=0)\nexpression_df.iloc[0:5,0:5]\n```\n\n| gene_name | S-1 | S-2 | S-3 | S-4 | S-5 |\n|-----------|--------|--------|--------|--------|--------|\n| MT-CO1 | 13.852 | 12.328 | 13.055 | 11.898 | 10.234 |\n| MT-CO2 | 13.406 | 12.383 | 13.281 | 11.578 | 11.156 |\n| MT-CO3 | 13.234 | 12.109 | 13.352 | 11.531 | 10.422 |\n| MT-ATP8 | 13.805 | 11.789 | 13.414 | 11.883 | 11.141 |\n| MT-ATP6 | 13.500 | 11.703 | 13.227 | 11.219 | 10.836 |\n\n```python\nXV = TSNE(n_components=2).\\\n fit_transform(expression_df.T)\ndf = pd.DataFrame(XV).rename(columns={0:'x',1:'y'})\n(ggplot(df,aes(x='x',y='y'))\n + geom_point(alpha=0.2)\n)\n```\n\n\n\nThe default command runs without verbose message output. but take notice, that genes that are not part of the `expression_df` are dropped from the analysis, and depending on your choice of GSVA method, genes for which there is not enough expression (i.e. all zero expression) will be dropped.\n\n```python\npathways_df = gsva(expression_df,genesets_df)\npathways_df.iloc[0:5,0:5]\n```\n\n| name | S-1 | S-2 | S-3 | S-4 | S-5 |\n|-------------------------|-----------|-----------|-----------|----------|-----------|\n| BIOCARTA_41BB_PATHWAY | 0.068631 | 0.257169 | -0.146907 | 0.020151 | -0.234537 |\n| BIOCARTA_ACE2_PATHWAY | 0.110822 | -0.222310 | -0.161572 | 0.370659 | -0.003318 |\n| BIOCARTA_ACH_PATHWAY | 0.514193 | 0.149291 | 0.226279 | 0.289960 | 0.016071 |\n| BIOCARTA_ACTINY_PATHWAY | -0.014494 | 0.407871 | -0.062163 | 0.055607 | 0.424726 |\n| BIOCARTA_AGPCR_PATHWAY | 0.622482 | -0.012845 | 0.317349 | 0.286368 | 0.022540 |\n\n```python\nYV = TSNE(n_components=2).\\\n fit_transform(pathways_df.T)\npf = pd.DataFrame(YV).rename(columns={0:'x',1:'y'})\n(ggplot(pf,aes(x='x',y='y'))\n + geom_point(alpha=0.2)\n)\n```\n\n\n\n\n",
"bugtrack_url": null,
"license": "Apache License, Version 2.0",
"summary": "Python CLI and module for running the GSVA R bioconductor package with Python Pandas inputs and outputs.",
"version": "1.0.6",
"split_keywords": [
"bioinformatics",
"r",
"enrichment",
"gsva",
"ssgsea",
"gsea",
"bioconductor"
],
"urls": [
{
"comment_text": "",
"digests": {
"md5": "788dec166139b75be041b339428854c0",
"sha256": "c655593bfcd7b1016fcfbfa9cf681ebbabf1744e856ae11a877ebfba1737e71b"
},
"downloads": -1,
"filename": "GSVA-1.0.6-py3-none-any.whl",
"has_sig": false,
"md5_digest": "788dec166139b75be041b339428854c0",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": null,
"size": 18445,
"upload_time": "2018-07-08T15:22:47",
"upload_time_iso_8601": "2018-07-08T15:22:47.241077Z",
"url": "https://files.pythonhosted.org/packages/c9/5b/dea37b86c9ac442579f7acdf538d04b077b81cf24da8455f204991e9882c/GSVA-1.0.6-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"md5": "fad6de38a625e1af1d367219c843d998",
"sha256": "4131f839fa29a13ef61dd191211361fcb6d53f0b64475a6fde82ec0c25472752"
},
"downloads": -1,
"filename": "GSVA-1.0.6.tar.gz",
"has_sig": false,
"md5_digest": "fad6de38a625e1af1d367219c843d998",
"packagetype": "sdist",
"python_version": "source",
"requires_python": null,
"size": 12301,
"upload_time": "2018-07-08T15:22:48",
"upload_time_iso_8601": "2018-07-08T15:22:48.500963Z",
"url": "https://files.pythonhosted.org/packages/e5/64/75cbd888bbd1574a16f89a49bdaf66d394c3d088b40929a334179d51770f/GSVA-1.0.6.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2018-07-08 15:22:48",
"github": true,
"gitlab": false,
"bitbucket": false,
"github_user": "jason-weirather",
"github_project": "GSVA",
"travis_ci": false,
"coveralls": false,
"github_actions": false,
"lcname": "gsva"
}
JSON
