RNA-seq data normalization in Python

Python implementation of common RNA-seq normalization methods:

CPM (Counts per million)
FPKM (Fragments per kilobase million)
TPM (Transcripts per million)
UQ (Upper quartile)
CUF (Counts adjusted with UQ factors)
TMM (Trimmed mean of M-values)
CTF (Counts adjusted with TMM factors)

For in-depth description of methods see documentation.

Features

Pure Python implementation (no need for R, etc.)
Compatible with Scikit-learn
Command line interface
Verbose documentation
Validated method implementation

Install

We recommend installing RNAnorm with pip:

pip install rnanorm

Quick start

The implemented methods can be executed from Python or from the command line.

Normalize from Python

The most common use case is to run normalization from Python:

>>> from rnanorm.datasets import load_toy_data
>>> from rnanorm import FPKM
>>> dataset = load_toy_data()
>>> # Expressions need to have genes in columns and samples in rows
>>> dataset.exp
          Gene_1  Gene_2  Gene_3  Gene_4  Gene_5
Sample_1     200     300     500    2000    7000
Sample_2     400     600    1000    4000   14000
Sample_3     200     300     500    2000   17000
Sample_4     200     300     500    2000    2000
>>> fpkm = FPKM(dataset.gtf_path).set_output(transform="pandas")
>>> fpkm.fit_transform(dataset.exp)
             Gene_1    Gene_2    Gene_3    Gene_4    Gene_5
Sample_1   100000.0  100000.0  100000.0  200000.0  700000.0
Sample_2   100000.0  100000.0  100000.0  200000.0  700000.0
Sample_3    50000.0   50000.0   50000.0  100000.0  850000.0
Sample_4   200000.0  200000.0  200000.0  400000.0  400000.0

Normalize from command line

Normalization from the command line is also supported. To list available methods and general help:

rnanorm --help

Get info about a particular method, e.g., CPM:

rnanorm cpm --help

To normalize with CPM:

rnanorm cpm exp.csv --out exp_cpm.csv

File exp.csv needs to be comma separated file with genes in columns and samples in rows. Values should be raw counts. The output is saved to exp_cpm.csv. Example of input file:

cat exp.csv
,Gene_1,Gene_2,Gene_3,Gene_4,Gene_5
Sample_1,200,300,500,2000,7000
Sample_2,400,600,1000,4000,14000
Sample_3,200,300,500,2000,17000
Sample_4,200,300,500,2000,2000

One can also provide input through standard input:

cat exp.csv | rnanorm cpm --out exp_cpm.csv

If file specified with --out already exists the command will fail. If you are sure that you wish to overwrite, use --force flag:

cat exp.csv | rnanorm cpm --force --out exp_cpm.csv

If no file is specified with --out parameter, output is printed to standard output:

cat exp.csv | rnanorm cpm > exp_cpm.csv

Methods TPM and FPKM require gene lengths. These can be provided either with GTF file or with "gene lengths" file. The later is a two columns file. The first column should include the genes in the header of exp.csv and the second column should contain gene lengths computed by union exon model:

# Use GTF file
rnanorm tpm exp.csv --gtf annotations.gtf > exp_out.csv
# Use gene lengths file
rnanorm tpm exp.csv --gene-lengths lenghts.csv > exp_out.csv
# Example of gene lengths file
cat lenghts.csv
gene_id,gene_length
Gene_1,200
Gene_2,300
Gene_3,500
Gene_4,1000
Gene_5,1000