Multi-substrate input
Natively processes DNA, RNA, and protein sequences. Protein inputs are reverse-mapped to nucleotides via an integrated tblastn search against NCBI.
DNARNAProtein → nt
§ 01
Built for rigorous miRNA discovery.
Six tightly-scoped capabilities, no glue code. Drop a target sequence in, get classified interactions back.
Strict & relaxed modes
Toggle between canonical exact seeds and a flexible wobble-pair search with re-weighted biological confidence.
--mode strict--mode relaxed
Thermodynamic scoring
Powered by ViennaRNA's RNAduplex to compute minimum free energy (MFE) of every candidate duplex.
ΔG kcal/mol
Confidence tiers
Every hit is graded — Very High, High, Medium, Low — using MFE, motif identity, AU context and cluster density.
4 tiers
Targeted validation
Skip whole-database scans. Hand the validator a list of miRNA IDs and get a clean YES/NO summary.
inline.txtFASTA
Flexible output
Switch between terminal-friendly summaries, full raw interaction tables, high-confidence-only filters, and direct CSV export — at any time, with one flag.
--output raw--output highconf--out results.csv
§ 02
Installation in three steps.
ViennaRNA can't be installed via pip, so we recommend a Conda environment. The whole thing takes about a minute.
01
Create and activate a Conda environment
Pin Python 3.11 — that's what ViennaRNA's pre-built bioconda channel targets.
~ shell
# Create the env
conda create -n mirnaprotpred_env python=3.11 -y
conda activate mirnaprotpred_env02
Install ViennaRNA
The structural-stability scorer. Bioconda ships everything you need; no compiler required.
~ shell
conda install -c bioconda viennarna -y03
Install mirnaprotpred from PyPI
That's it — both SeqFinder and validator become available on your PATH.
~ shell
pip install mirnaprotpred§ 03
Two engines, one biological model.
Use SeqFinder to discover every potential interaction in a target. Use validator when you already know which miRNAs you care about.
Scan an entire sequence — FASTA file or inline string — and surface all potential miRNA interactions, ranked by composite confidence.
$SeqFinder<target>[--mode][--output][--out]
▸ Scan a genome
Drop in a FASTA file. Defaults to strict mode and a concise summary.
SeqFinder examples/sars.fasta
▸ Include wobble pairs
Relaxed mode catches G·U wobbles and re-weights confidence accordingly.
SeqFinder examples/sars.fasta --mode relaxed
▸ Full raw output
Every interaction, every metric — MFE, AU context, motif, cluster, the lot.
SeqFinder examples/sars.fasta --output raw
▸ High-confidence only
Filter down to High & Very High tier hits for quick review.
SeqFinder examples/sars.fasta --output highconf
▸ Export to CSV
Combine any output mode with --out to write a CSV alongside.
SeqFinder sars.fasta --output raw --out results.csv
▸ Inline sequence
No file needed — pass a quoted nucleotide string directly.
SeqFinder "AUGCAUGCAUGCAUGC"
§ 04
The engine, stage by stage.
Every input passes through the same six-stage biological pipeline before a confidence grade is emitted.
01
Input detection
Auto-detects DNA, RNA or protein. Protein inputs trigger a tblastn search to recover the source nucleotide region.
nt · aa ·tblastn
02
Motif discovery
Boyer–Moore string matching locates every canonical or relaxed seed variant across the target at speed.
seeds ·Boyer–Moore
03
Contextual scoring
The 20 nt flanking each hit is scored for AU content, a proxy for structural accessibility.
±20 nt ·AU-rich
04
Structural analysis
RNAduplex computes the optimal secondary structure & minimum free energy of the miRNA–target duplex.
ΔG kcal/mol ·ViennaRNA
05
Clustering & ranking
Competing seeds at the same locus are clustered and the optimal interaction wins on composite score.
composite ·winner-take-locus
06
Confidence grading
Hits land in one of four tiers based on stringent thresholds across all four scoring axes.
Very High · High · Med · Low
§ 05
Contact& citation.
Bug reports, feature requests, citation questions — drop the team a line directly.
“If you use miRNAProtPred in your research, please cite us.”
Citation details pending publication — full BibTeX will appear here on release.