This is an example of how to prepare the input file(s) and run metilene3. You should have a table of your samples and a folder with BED files. Here, we use Python3 and bedtools.

import os
import pandas as pd

Load the table that contains sample IDs, group information (if you want to use supervised mode) and names of BED files:

df = pd.read_excel('samples.xlsx')
df[['ID','Group']].to_csv("group.tsv", sep='\t', index=False,)
df[['ID','Group','bedfile']].head(5)

	ID	Group	bedfile
0	S0	Blood-T	S0.bed
1	S100	Blood-B	S100.bed
2	S101	Blood-NK	S101.bed
3	S102	Blood-T	S102.bed
4	S103	Blood-T	S103.bed

Use bedtools to merge all BED files to a matrix:

os.system('cd /path_to_BEDfiles/;\
bedtools unionbedg -i '+(df['bedfile']+' ').sum()+\
'-header -names '+(df['ID']+' ').sum()+\
'-filler NA > ./input.raw')

Load the merged matrix and convert it to the metilene3 format:

met = pd.read_table('./input.raw')
met = met.drop(columns='start')
met.to_csv("./input.tsv", sep='\t', index=False, float_format='%.3f', na_rep='.')
met.head(5)

	chrom	end	S0	S100	S101	S102	S103	S104	S105	S106	...	S91	S92	S93	S94	S95	S96	S97	S98	S99	S9
0	chr1	10470	0.666667	NaN	0.675000	0.625000	0.575000	0.652174	0.760000	0.552632	...	0.690476	0.700000	0.562500	0.708333	0.875000	0.700000	0.968750	0.550000	0.727273	0.774194
1	chr1	10472	0.600000	0.900000	0.725000	0.800000	0.846154	0.869565	0.750000	0.625000	...	0.880952	0.750000	0.740000	0.863636	0.833333	0.633333	0.968750	0.625000	0.909091	0.882353
2	chr1	10485	0.875000	0.909091	0.682927	0.904762	0.790698	0.846154	0.814815	0.727273	...	0.886364	0.818182	0.758621	0.800000	0.958333	0.812500	0.976744	0.795918	0.962963	0.897436
3	chr1	10490	0.882353	1.000000	0.860465	1.000000	0.795455	0.827586	0.964286	0.767442	...	0.866667	0.878788	0.896552	0.920000	0.875000	0.787879	1.000000	0.843137	0.896552	0.925000
4	chr1	10494	0.705882	0.818182	0.813953	1.000000	0.727273	0.678571	0.586207	0.642857	...	0.911111	0.764706	0.766667	0.760000	0.791667	0.848485	0.979167	0.673077	0.933333	0.875000

Run metilene3 with unsupervised mode:

os.system('python /path_to_metilene3/metilene3.py \
  -i ./input.tsv \
  -o ./results_unsupervised')

Or, run metilene3 with supervised mode:

os.system('python /path_to_metilene3/metilene3.py \
  -i ./input.tsv \
  -g ./group.tsv \
  -o ./results_supervised')

Finally you can find the results under ./results_unsupervised or ./results_supervised.