Documentation:

https://multitraitclustering.readthedocs.io

Requirements:

• python == 3.11.9 (not currently compatible with higher versions)
• scipy == 1.15.1
• scikit-learn == 1.6.1
• scikit-learn-extra == 0.3.0
• numpy == 2.2.2
• pandas == 2.2.3

Setup:

Installation:

pip install multitraitclustering

Data:

• Association Data. This is your master data set with association scores corresponding

to a collection of GWAS analysis. The columns correspond to traits, the rows and snps and the values are the association between the snp and the trait. This should be stored as a .csv file with the trait labels on the first row and the snps on the first column. * Exposure Data. This is a dataset corresponding to only the exposure phenotype and the association with the snps. Also stored as a csv file, rows are snps, one column for the exposure labelled EXP and the values are the scores. * P-values and Standard Error values are not yet incorporated into the analysis.

Usage Example:

Loading Data

To load data located in a directory named data whose parent directory is your working directory. The filenames eff_dfs.csv and exp_dfs.csv can be replaced with alternative filenames to match your files. The files must be in csv format.

from multitraitclustering import data_setup as ds
data_sets = ds.load_association_data("./data",
                                    eff_fname ="/eff_dfs.csv",
                                    exp_fname = "/exp_dfs.csv")

This will give an output in the form of:

data_sets = {'eff_df':X1687     X1697     X1717        X20015_irnt  X21001_irnt
rs115866895  0.086834 -0.970112 -1.105207     0.668033     0.402944
rs4648450   -0.070006 -0.211097 -1.109551     0.114491    -0.253709
rs12024554   0.202266  0.700323 -0.281241     3.877386    -0.650558
rs1097327   -0.047729 -0.370089 -0.148402     0.550493    -0.659329
rs3737992   -0.741344  1.597846 -1.133015     1.346192    -0.489776   ,
'exp_df':   EXP
rs115866895 -0.013280
rs4648450   -0.010908
rs12024554  -0.009474
rs1097327   -0.009443
rs3737992   -0.010055 }

Clustering analysis

from multitraitclustering import data_setup as ds
from multitraitclustering import multi_trait_clustering as mtc

data_sets = ds.load_association_data("./data",
                                eff_fname ="/eff_dfs.csv",
                                exp_fname = "/exp_dfs.csv")

cluster_res = mtc.cluster_all_methods(data_sets["eff_df"], data_sets["exp_df"])

The results contains the clusters and the clustering parameters clust_pars_dict and clust_results.

clust_pars = cluster_res["clust_pars_dict"]
clust_df = cluster_res["clust_results"]