sumi: A simple small RNA umi analysis

A simple analysis for small RNA libraries with UMIs

Can be used to check for the presence of isomiRs in addition to canonical miRNAs.

It performs UMI error correction and deduplication using a directional graph algorithm. This script implements a slightly modified directional graph algorithm that allows for a Hamming Distance of 1 between UMIs, see Fu, Y., et al, (2018). Elimination of PCR duplicates in RNA-seq and small RNA-seq using unique molecular identifiers. https://doi.org/10.1186/s12864-018-4933-1. It uses a five fold threshold, while the original algorithm uses a two fold count threshold.

The UMI deduplication process is multithreaded to significantly speed up the process.

Script dependencies

• bowtie2
• samtools
• cutadapt

Installation instructions

Create executable to run on local machine using the compiled binary.

You will need to have Rust installed on your computer before starting. Rust can be installed from here: Install Rust

Install the binary using cargo.

cargo install sumi

The compiled binary will be located in ~/.cargo/bin/. Make it executable with the following command:

chmod +x ~/.cargo/bin/sumi 

There are numerous options that can be changed if desired. Use -h or --help flags to see options.

./sumi --help

Basic usage

The app can be run using the sumi executable. To this command to analyze files with a miRNA bowtie2 reference located in /home/user/data/, a 12 bp UMI on the 5' end with the structure "NNNNNCCANNTCANNNNN", and with 8 threads.

cd fastqs
./sumi --reference /home/user/data/miRNA --umi-regex "^(.{5})CCA(.{2})TCA(.{5})" --threads 8

To check for isomiRs and generate counts and the alignment information, run the following command. Note that the isomiR and canonical miRNA percentages are separated.

./sumi --reference /home/user/data/miRNA --isomir --write-metrics

This can be used to analyze libraries with a 12 bp 3' UMI with a 1 bp mismatch allowed during alignment. Make sure the regex pattern contains an anchor, such as '$', so that it is specific to the 3' end.

./sumi --reference /home/user/data/miRNA --3p --umi-regex "(.{12}$)" --mismatch

For analyzing isomiRs in Qiagen libraries use this command, there is no need to specify the UMI is on the 3' end or regex pattern.

./sumi --reference /home/user/data/miRNA --isomir --qiagen