PALS

Logo

A Python tool to rank significantly-changing metabolite sets

View the Project on GitHub glasgowcompbio/PALS

1. Introduction

To access our interactive Web application PALS Viewer, please visit https://pals.glasgowcompbio.org/app/.

Pathway analysis is an important task in understanding complex metabolomic data. Here we introduce PALS (Pathway Activity Level Scoring), a complete tool that performs database queries of pathways, decomposes activity levels in pathways via the PLAGE method, as well as presents the results in a user-friendly manner. The results are found to be more robust to noise and missing peaks compared to the alternatives (ORA, GSEA). This is particularly important for metabolomics peak data, where noise and missing peaks are prevalent.

PALS

Additionally, the decomposition approach in PALS is amenable to the analysis of any group of metabolite sets, not just pathways. As demonstrated in PALS Viewer, metabolite sets obtained from the grouping of metabolites according to their fragmentation spectra can also be analysed. This includes in particular Molecular Families from GNPS, as well as Mass2Motifs from MS2LDA. From PALS Viewer, you can also prioritise MF and Mass2Motifs from your GNPS analysis based on their activity levels. For more details including how to analyse user-defined metabolite sets from Jupyter notebooks, see Section 8.

2. Installation

For the latest development version, check out this repository using Git. Otherwise PALS can also be installed via pip install PALS-pathway.

To use Reactome as pathway database, refer to the setup guide.

3. Running PALS

To run PALS from the command-line, the script pals/run_pals.py is used. This script accepts a number of parameters, documented below (bold indicates required parameters).

Note: if you have installed PALS via pip, then run_pals.py (Unix-based systems) or run_pals.exe (Windows) will also be added to your path during installation. It can be run directly by typing its name from the shell.

Usage Parameters

usage: run.py [-h] --db {PiMP_KEGG,COMPOUND,ChEBI,UniProt,ENSEMBL}
              --comparisons COMPARISONS [COMPARISONS ...]
              [--min_replace MIN_REPLACE]
              [--species {Arabidopsis thaliana,Bos taurus,Caenorhabditis elegans,
              Canis lupus familiaris,Danio rerio,Dictyostelium discoideum,
              Drosophila melanogaster,Gallus gallus,Homo sapiens,Mus musculus,
              Oryza sativa,Rattus norvegicus,Saccharomyces cerevisiae,Sus scrofa}]
              [--use_all_reactome_pathways] [--connect_to_reactome_server]
              {PLAGE,ORA,GSEA} intensity_csv annotation_csv output_file

Command-line Examples

The most basic usage of PALS is to run it in offline-mode using PLAGE as the decomposition method. This uses the downloaded KEGG database for pathways. Here we run PALS on the example HAT data used in the manuscript

$ python pals/run.py PLAGE notebooks/test_data/HAT/int_df.csv notebooks/test_data/HAT/annotation_df.csv test_output.csv --db PiMP_KEGG --comparisons Stage_1/Control Stage_2/Control

Downloaded Reactome pathways is also provided in PALS for the most common species. Note that only metabolic pathways are available in this mode. Below shows an example run using Human pathways.

$ python pals/run.py PLAGE notebooks/test_data/HAT/int_df.csv notebooks/test_data/HAT/annotation_df.csv test_output.csv --db COMPOUND --comparisons Stage_1/Control Stage_2/Control --min_replace 5000 --species "Homo sapiens"

Finally in online mode, PALS can connect to a Reactome database instance to retrieve the most updated pathways. The flag --use_all_reactome_pathways specifies that all pathways should be used (not just metabolic pathways), while the flag --connect_to_reactome_server defines that online mode should be used.

$ python pals/run.py PLAGE notebooks/test_data/HAT/int_df.csv notebooks/test_data/HAT/annotation_df.csv test_output.csv --db COMPOUND --comparisons Stage_1/Control Stage_2/Control --min_replace 5000 --species "Homo sapiens" --use_all_reactome_pathways --connect_to_reactome_server

The following example output is produced:

output

Pathways are identified by their id and can be sorted by the p-value columns. The column unq_pw_F lists the unique formulae found in that pathway, tot_ds_F lists the formula hits found in the dataset, and F_coverage is the proportion of tot_ds_F to unq_pw_F.

4. File format and data imputation

Users provide two input files to PALS. The first is a matrix is of individual peak intensities (rows are peak features with column one containing the peak id, further columns representing individual samples). If using the CSV file, the second line can be used to indicate which groups this sample belongs to. For example, the intensity matrix takes the form of:

row_id,Beer_1_full1.mzXML,Beer_1_full2.mzXML,Beer_1_full3.mzXML,
    Beer_2_full1.mzXML,Beer_2_full2.mzXML,Beer_2_full3.mzXML,
    Beer_3_full1.mzXML,Beer_3_full2.mzXML,Beer_3_full3.mzXML,
    Beer_4_full1.mzXML,Beer_4_full2.mzXML,Beer_4_full3.mzXML
group,beer1,beer1,beer1,beer2,beer2,beer2,beer3,beer3,beer3,beer4,beer4,beer4
3033929,2235291136,2000478208,2170697216,2242759936,2279881984,1959479680,
    2079356160,2110473216,2243652608,1817064704,1746442752,1779827200
3033930,44334908,42873872,48948532,47604480,42172796,39084524,
    38257776,37701920,40871888,33304766,31536296,31024098

Data imputation is performed to the intensity matrix when it is loaded: if all of the samples in a single experimental factor have intensities of zero these are replaced by the minimum intensity value (which can be set by the user); and if only some of the sample values in a factor are zero then these are replaced by the mean value of the non-zero samples in that factor. The data is subsequently transformed to log-2 base and standardised using the preprocessing module in Scipy such that the intensity matrix has a zero mean and unit variance across the samples.

In addition, users also provide a list of compound annotations assigned to peak features (peaks that do not have annotations will not be used for pathway analysis). As a result of the uncertainty in peak identification, multiple peak IDs may be mapped to multiple compound IDs and \textit{vice versa}. As such, annotations are provided as another matrix having two columns. The first column (or DataFrame index) is the peak ID while the second column is the assigned metabolite annotation as either KEGG or ChEBI database IDs.

row_id,entity_id
3033929,C00148
3033930,C06326
3033931,C00183
3033931,C00719
3033931,C00583

5. Other Pathway Analysis Methods

ORA and GSEA are used as comparisons and are both included in PALS, and can be used for benchmarking and analysis. For more details, please refer to our paper.

6. Web Interface

PALS Viewer is a Web interface on top of the Streamlit framework. It can be used to run PALS, analyse pathway ranking results as well as inspect significantly changing pathways. To run it locally use following command:

$ streamlit run pals/run_gui.py

Alternatively an instance of PALS Viewer can also be found on our server.

7. Importing PALS for use in other applications

PALS can be imported as a Python library and incorporated into your own Python application. This is illustrated in the following code snippet:

from pals.ORA import ORA
from pals.PLAGE import PLAGE
from pals.GSEA import GSEA
from pals.common import *
from pals.feature_extraction import DataSource

# TODO: correctly initialise the following data structures for your data.
# Refer to the paragraphs below.
int_df = pd.DataFrame()
annotation_df = pd.DataFrame()
experimental_design = {}

# Using Reactome pathways matching by KEGG ID
database_name = 'COMPOUND'

# If true, we limit to metabolic pathways only. Otherwise all pathways will be queried.
reactome_metabolic_pathway_only = True

# If true, we use online mode that queries Reactome on a local Neo4j server. 
# Otherwise offline mode will be used (using downloaded database files).
reactome_query = True

# Minimum intensity value for data imputation
min_replace = 5000

ds = DataSource(int_df, annotation_df, experimental_design, database_name,
                reactome_species=reactome_species,
                reactome_metabolic_pathway_only=reactome_metabolic_pathway_only,
                reactome_query=reactome_query, min_replace=min_replace)

method = PLAGE(ds)
# method = ORA(ds) 
# method = GSEA(ds)

df = method.get_pathway_df()

When PALS is used programatically, pandas dataframes storing relevant data can be passed directly. Experimental design data can be passed directly as a dictionary structure in the programmatic use. In the example above, int_df is the intensity data frame containing peak intensity information described in the File Format section above (with the second line of grouping information omitted). Similarly annot_df is the annotation data frame containing peak annotations also described above.

Example int_df dataframe: int_df

Example annot_df dataframe:

The experimental design data in experimental_design contains information on groups, which relates all samples in a particular experimental factor together as well as comparisons, which describes the desired comparisons for the PALS analysis in terms of a case and a control. An example of this can be found below:

experimental_design = {
    'comparisons': [
        {'case': 'beer1', 'control': 'beer2', 'name':  'beer1/beer2'},
        {'case': 'beer3', 'control': 'beer4', 'name': 'beer3/beer4'}
    ],
    'groups': {
        'beer1': [
            'Beer_1_full2.mzXML', 
            'Beer_1_full1.mzXML',
            'Beer_1_full3.mzXML'
        ],
        'beer2': [
            'Beer_2_full3.mzXML', 
            'Beer_2_full1.mzXML', 
            'Beer_2_full2.mzXML'
        ],
        'beer3': [
            'Beer_3_full3.mzXML', 
            'Beer_3_full2.mzXML', 
            'Beer_3_full1.mzXML'
        ],
        'beer4': [
            'Beer_4_full3.mzXML',
            'Beer_4_full2.mzXML',
            'Beer_4_full1.mzXML'
        ],
    }
}

Many example notebooks are provided in the notebooks folders. If you have any question in using PALS in your application, please raise an issue or simply email us.’

8. Analysis of User-defined Metabolite Sets

PALS can be used to analyse any user-defined group of metabolites. In the paper, we demonstrated this through an example by analysing potentially unknown metabolites that have been grouped into Molecular Families (by GNPS) and into Mass2Motifs (by MS2LDA). The following notebook demonstrates how the analysis was done. Additionally Molecular Families and Mass2Motifs analysis have also been included as an option in PALS Viewer.

9. Publication

If you are using PALS, please cite the following publication.

Mcluskey, K., Wandy, J., Vincent, I., Hooft, J. J. J. Van Der, Rogers, S., Burgess, K. & Daly, R. (2021). Ranking Metabolite Sets by Their Activity Levels. Metabolites. 11(2):103 https://doi.org/10.3390/metabo11020103