Skip to content

Database download and setup guide for sRNA-IMP

This bundle documents and automates setup of the major databases used by sRNA-IMP.

Covered resources

Human arm

  • Human miRNA reference
  • Human tRNA reference
  • Human rRNA reference
  • Human other ncRNA reference

Non-human arm

  • RiboDetector model setup (tool installation, no large external DB required)
  • Kraken2 standard taxonomy database
  • Bracken k-mer distribution files
  • Rfam covariance models
  • Optional non-human miRNA references
  • Optional combined tRNA references

Novel ncRNA discovery

  • Rfam covariance models
  • ARAGORN / tRNAscan-SE installation via conda environment
databases/
├── human/
│   ├── mirna/
│   ├── trna/
│   ├── rrna/
│   └── otherrna/
├── nonhuman/
│   ├── kraken2/
│   ├── bracken/
│   ├── mirna/
│   ├── trna/
│   └── rfam/
└── shared/
    └── rfam/

Quick-start

  1. Create the conda / micromamba environment that contains the required tools.
  2. Edit config_examples/database_paths.example.sh
  3. Run: 
    bash scripts/download_and_setup_databases.sh

Notes on data sources

Rfam

The official Rfam genome-annotation guide recommends downloading Rfam.cm.gz and Rfam.clanin, then indexing Rfam.cm with cmpress. Use cmscan --rfam --cut_ga for annotation against these models.

Kraken2

The official Kraken2 manual documents kraken2-build --standard --db <DB> for building the standard database.

Bracken

Bracken requires a built Kraken database plus generation of the database*kmer_distrib files with bracken-build.

tRNAscan-SE

tRNAscan-SE 2.0 is available from the UCSC Lowe Lab GitHub and website. It depends on Infernal.

ARAGORN

ARAGORN is also available through Bioconda, which is the simplest reproducible installation route for pipeline use.

RiboDetector

RiboDetector is distributed via Bioconda and GitHub. It is a tool rather than a large downloadable reference database.

miRNA references

For miRNA references, use either: - miRBase (mature.fa, hairpin.fa) when you want broad species coverage - MirGeneDB when you want a more conservative, curated animal miRNA set

Human arm reference recommendations

Human miRNA

Use one of: - miRBase mature.fa filtered to hsa-* - MirGeneDB mature sequences for human

Build a Bowtie index: 

bowtie-build hsa_mature.fa hsa_mature

Human tRNA

Preferred sources: - GtRNAdb exports for human nuclear tRNAs - human mitochondrial tRNAs from a curated FASTA

Build: 

cat human_tRNA.fa human_mt_tRNA.fa > hsa_tRNA_all.fa
bowtie-build hsa_tRNA_all.fa hsa_tRNA_all

Human rRNA

Use curated human cytosolic + mitochondrial rRNA FASTA, then build: 

bowtie-build hsa_rRNA.fa hsa_rRNA

Human other ncRNA

A practical source is Ensembl ncRNA FASTA: 

bowtie-build Homo_sapiens.GRCh38.ncrna.fa hsa_other_ncRNA

Non-human arm reference recommendations

Kraken2 + Bracken

  • Build Kraken2 standard DB
  • Build Bracken files for your read length (50 bp typical for your pipeline)

Rfam

Use the full Rfam.cm, then optionally extract a smaller subset later if runtime becomes an issue.

Optional non-human miRNA reference

Use miRBase mature.fa, exclude hsa, and optionally subset to: - plants - fungi - protozoa - viruses

Optional tRNA reference

For broader non-human tRNA screening, create a combined FASTA containing: - bacterial tRNAs - fungal / eukaryotic tRNAs - plant chloroplast tRNAs - plant mitochondrial tRNAs

Runtime / storage expectations

Resource Approximate size Comment
Kraken2 standard DB ~100 GB during build largest resource
Rfam.cm + indices several GB shared across workflows
Bracken files depends on Kraken DB build once per DB
Bowtie miRNA/tRNA/rRNA refs MB to low GB lightweight

Verification checklist

After setup, verify: - cmpress generated Rfam.cm.i1f, i1i, i1m, i1p - Kraken2 DB contains hash.k2d, opts.k2d, taxo.k2d - Bracken DB contains database50mers.kmer_distrib (or matching read length) - each Bowtie index has .1.ebwt ... .4.ebwt files

Suggested provenance file

Create a plain text manifest such as: 

Rfam=15.1
Kraken2_standard_built=2026-03-27
Bracken_read_length=50
miRBase=22.1
MirGeneDB_downloaded=2026-03-27