sspa 1.0.4

Single sample pathway analysis tools for omics data

sspa

Single sample pathway analysis toolkit

sspa provides a Python interface for metabolomics pathway analysis. In addition to conventional methods over-representation analysis (ORA) and gene/metabolite set enrichment analysis (GSEA), it also provides a wide range of single-sample pathway analysis (ssPA) methods.

Features

• Over-representation analysis
• Metabolite set enrichment analysis (based on GSEA)
• Single-sample pathway analysis
• Compound identifier conversion
• Pathway database download (KEGG, Reactome, and PathBank)

Although this package is designed to provide a user-friendly interface for metabolomics pathway analysis, the methods are also applicable to other datatypes such as normalised RNA-seq and proteomics data.

Documentation and tutorials

This README provides a quickstart guide to the package and its functions. For new users we highly recommend following our full walkthrough notebook tutorial available on Google Colab which provides a step-by-step guide to using the package.

Click the link above and save a copy of the Colab notebook to your Google Drive. Alternatively, you can download the notebook from the Colab tutorial as an '.ipynb' file and run it locally using Jupyter Notebook or Jupyter Lab.

Documentation is available on our Read the Docs page. This includes a function API reference.

Quickstart

pip install sspa

Load Reactome pathways

reactome_pathways  = sspa.process_reactome(organism="Homo sapiens")

Load some example metabolomics data in the form of a pandas DataFrame:

covid_data_processed = sspa.load_example_data(omicstype="metabolomics", processed=True)

Generate pathway scores using kPCA method

kpca_scores = sspa.sspa_kpca(reactome_pathways, min_entity=2).fit_transform(covid_data_processed.iloc[:, :-2])

Loading example data

Note we provide processed and non-processed versins of the COVID example metabolomics dataset (Su et al. 2020, Cell). The processed version (set processed=True) already has ChEBI identifiers as column names, whereas the non-processed version has metabolite names.

covid_data = sspa.load_example_data(omicstype="metabolomics", processed=False)

Here we demonstrate some simple pre-processing for this dataset in order to enable conventional and ssPA pathway analysis:

# Keep only metabolites (exclude metadata columns)
covid_values = covid_data.iloc[:, :-2]

# Remove metabolites with too many NA values
data_filt = covid_values.loc[:, covid_values.isin([' ', np.nan, 0]).mean() < 0.5]

# Impute using the median
imputed_mat = data_filt.fillna(data_filt.median())

# Log transform the data
log2_mat = np.log2(imputed_mat)

# Standardise the data (metabolite values) using z-score (mean=0, std=1) by subtracting the mean and dividing by the standard deviation
processed_data = (log2_mat - log2_mat.mean(axis=0)) / log2_mat.std(axis=0)

Loading pathways

# Pre-loaded pathways
# Reactome v78
reactome_pathways = sspa.process_reactome(organism="Homo sapiens")

# KEGG v98
kegg_human_pathways = sspa.process_kegg(organism="hsa")

Load a custom GMT file (extension .gmt or .csv)

custom_pathways = sspa.process_gmt("wikipathways-20220310-gmt-Homo_sapiens.gmt")

Download latest version of pathways

# download KEGG latest metabolomics pathways
kegg_mouse_latest = sspa.process_kegg("mmu", download_latest=True, filepath=".")

# download Reactome latest metabolomics pathways
reactome_mouse_latest = sspa.process_reactome("Mus musculus", download_latest=True, filepath=".", omics_type='metabolomics')

# download Pathbank latest metabolomics pathways
pathbank_human_latest = sspa.process_pathbank("Homo sapiens", download_latest=True, filepath=".", omics_type='metabolomics')

Download latest version of multi-omics pathways

• For Reactome, users can specify the omics types required via the 'identifiers' argument. Leaving this to None downloads all omics (ChEBI, UniProt, Gene Symbol). Users can specify any combination of ['chebi', 'uniprot', 'gene_symbol'].
• For KEGG, multi-omics pathways are represented by KEGG gene and KEGG compound identifiers.

# download multi-omics pathways from Reactome (ChEBI, UniProt, Gene Symbol)
reactome_human_mo = sspa.process_reactome('Homo sapiens', download_latest=True, filepath=".", omics_type='multiomics', identifiers=['chebi', 'uniprot', 'gene_symbol'])

# download multi-omics pathways from Reactome (ChEBI and UniProt)
reactome_human_mo = sspa.process_reactome('Homo sapiens', download_latest=True, filepath=".", omics_type='multiomics', identifiers=['chebi', 'uniprot'])

# download multi-omics pathways from KEGG (KEGG gene and KEGG compound)
kegg_mouse_latest = sspa.process_kegg("mmu", download_latest=True, filepath=".", omics_type='multiomics')

Identifier harmonization

Note: KEGG pathways use KEGG compound IDs, Reactome and Pathbank pathways use ChEBI and UniProt (for proteins)

# download the conversion table
compound_names = processed_data.columns.tolist()
conversion_table = sspa.identifier_conversion(input_type="name", compound_list=compound_names)

# map the identifiers to your dataset
processed_data_mapped = sspa.map_identifiers(conversion_table, output_id_type="ChEBI", matrix=processed_data)

Conventional pathway analysis

Over-representation analysis (ORA)

ora = sspa.sspa_ora(processed_data_mapped, covid_data["Group"], reactome_pathways, 0.05, DA_testtype='ttest', custom_background=None)

# perform ORA 
ora_res = ora.over_representation_analysis()

# get t-test results
ora.ttest_res

# obtain list of differential molecules input to ORA
ora.DA_test_res

Gene Set Enrichment Analysis (GSEA), applicable to any type of omics data

sspa.sspa_gsea(processed_data_mapped, covid_data['Group'], reactome_pathways)

Single sample pathway analysis methods

All ssPA methods now have a fit(), transform() and fit_transform() method for compatibility with SciKitLearn. This allows integration of ssPA transformation with various machine learning functions in SKLearn such as Pipeline and GridSearchCV. Specifically for sspa.sspa_ssClustPA, sspa.sspa_SVD, and sspa.sspa_KPCA methods the model can be fit on the training data and the test data is transformed using the fitted model.

# ssclustPA
ssclustpa_res = sspa.sspa_ssClustPA(reactome_pathways, min_entity=2).fit_transform(processed_data_mapped)

# kPCA 
kpca_scores = sspa.sspa_kpca(reactome_pathways, min_entity=2).fit_transform(processed_data_mapped)

# z-score (Lee et al. 2008)
zscore_res = sspa.sspa_zscore(reactome_pathways, min_entity=2).fit_transform(processed_data_mapped)

# SVD (PLAGE, Tomfohr et al. 2005)
svd_res = sspa.sspa_svd(reactome_pathways, min_entity=2).fit_transform(processed_data_mapped)

# ssGSEA (Barbie et al. 2009)
ssgsea_res = sspa.sspa_ssGSEA(reactome_pathways, min_entity=2).fit_transform(processed_data_mapped)

License

GNU GPL 3.0

Citing us

If you found this package useful, please consider citing us:

ssPA package

@article{Wieder22a,
   author = {Cecilia Wieder and Nathalie Poupin and Clément Frainay and Florence Vinson and Juliette Cooke and Rachel PJ Lai and Jacob G Bundy and Fabien Jourdan and Timothy MD Ebbels},
   doi = {10.5281/ZENODO.6959120},
   month = {8},
   title = {cwieder/py-ssPA: v1.0.4},
   url = {https://zenodo.org/record/6959120},
   year = {2022},
}

Single-sample pathway analysis in metabolomics

@article{Wieder2022,
   author = {Cecilia Wieder and Rachel P J Lai and Timothy M D Ebbels},
   doi = {10.1186/s12859-022-05005-1},
   issn = {1471-2105},
   issue = {1},
   journal = {BMC Bioinformatics},
   pages = {481},
   title = {Single sample pathway analysis in metabolomics: performance evaluation and application},
   volume = {23},
   url = {https://doi.org/10.1186/s12859-022-05005-1},
   year = {2022},
}

Contributing

Read our contributor's guide to get started

Contributors

We are grateful for our contributors who help develop and maintain py-ssPA:

• Maëlick Brochut @mbrochut

News and updates

Read more

[v1.0.2] - 4/12/23

• Enable download of Pathbank pathways (metabolite and protein) via the process_pathbank() function

[v1.0.0] - 25/08/23

• Add compatability with SciKitLearn by implementing fit(), transform() and fit_transform() methods for all ssPA methods. This allows integration of ssPA transformation with various machine learning functions in SKLearn such as Pipeline and GridSearchCV. Specifically for sspa.sspa_ssClustPA, sspa.sspa_SVD, and sspa.sspa_KPCA methods the model can be fit on the training data and the test data is transformed using the fitted model.
• Fixed ID conversion bug in sspa.map_identifiers() due to MetaboAnalyst API URL change

[v0.2.4] - 04/07/23

Enable the download of multi-omics (ChEBI and UniProt) Reactome pathways for multi-omics integration purposes. Enable omics_type='multiomics' to download:

reactome_mouse_latest_mo = sspa.process_reactome("Mus musculus", download_latest=True, filepath=".", omics_type='multiomics')

[v0.2.3] - 23/06/23

• @mbrochut Bug fix in KEGG pathway downloader
• @mbrochut Add tqdm progress bar for long KEGG downloads

[v0.2.1] - 05/01/23

• Removal of rpy2 dependency for improved compatibility across systems
• Use GSEApy as backend for GSEA and ssGSEA
• Minor syntax changes
  • ora.ttest_res is now ora.DA_test_res (as we can implement t-test or MWU tests)
  • sspa_fgsea() is now sspa_gsea() and uses gseapy as the backend rather than R fgsea
  • sspa_gsva() is temporarily deprecated due to the need for the rpy2 compatability - use the GSVA R package