23. beta_topN

23.1. Overview

beta_topN ranks CpGs by a row-wise summary statistic and retains the top N features for downstream analyses such as PCA or clustering.

The input matrix should have CpGs in rows and samples in columns.

23.2. Ranking Scores

Supported row-wise scores are:

Score	Description
`std`	Standard deviation across samples. This is the default.
`var`	Variance across samples.
`mean`	Mean across samples.
`median`	Median across samples.
`mad`	Median absolute deviation across samples.

23.3. Input

The input is a tabular CpG-by-sample matrix. Delimiters are detected automatically and compressed input is supported.

Example:

CpG_ID   Sample_01   Sample_02   Sample_03   Sample_04
cg_001   0.831035    0.878022    0.794427    0.880911
cg_002   0.249544    0.209949    0.234294    0.236680
cg_003   0.845065    0.843957    0.840184    0.824286

23.4. Usage

Basic usage:

beta_topN \
    -i test_05_TwoGroup.tsv.gz \
    -c 500 \
    -o test_05_TwoGroup

Useful options include:

-c, --count, --top – number of CpGs to retain (default: 1000)
-s, --score – std, var, mean, median, or mad
--ascending – rank from smallest to largest score
--na_policy {drop,omit} – remove incomplete CpGs or calculate from available values
--min_valid – minimum observed values required when --na_policy omit is used (default: 2)
-o, --out_prefix, --output – output prefix

Display all options with:

beta_topN -h

23.5. Output

For output prefix test_05_TwoGroup, the command writes:

test_05_TwoGroup.ranked.tsv – all retained CpGs with Score and Rank columns
test_05_TwoGroup.topN.tsv – top-ranked CpG-by-sample matrix
test_05_TwoGroup.topN_ids.txt – selected CpG IDs

23.6. Example Data

test_05_TwoGroup.tsv.gz