miRanda

miRanda is a microRNA target-site prediction program written by Enright et al. at Memorial Sloan Kettering Cancer Center. It predicts where a microRNA binds an RNA or DNA target by combining a complementarity-based local alignment with a thermodynamic estimate of the resulting duplex’s stability, and returns each predicted site with its score, free energy, alignment, and coordinates. This makes it useful for nominating candidate microRNA targets in 3’UTRs and other transcript regions ahead of experimental validation.

Memorial Sloan Kettering Cancer Center

Proto is not affiliated with Memorial Sloan Kettering Cancer Center. This toolkit is open source and builds on the implementation produced by this organization. Product names, logos, and trademarks are the property of their respective owners.

---

GitHub 12 GitHub 12 Publication Publication Cite Cite Tool Source Tool Source Open as Notebook Open as Notebook

hacktrackgnulinux/miranda

[Pentest] miRanda: An miRNA target scanner that aims to predict mRNA targets for microRNAs using dynamic-programming alignment and thermodynamics

12 stars

View repo

MicroRNA targets in Drosophila

Anton J Enright, Bino John, … Debora S Marks

Genome Biology (2003)

Read paper

@article{enright2003microrna,
  title={MicroRNA targets in Drosophila},
  author={Enright, Anton J and John, Bino and Gaul, Ulrike and Tuschl, Thomas and Sander, Chris and Marks, Debora S},
  journal={Genome Biology},
  volume={5},
  number={1},
  pages={R1},
  year={2003},
  doi={10.1186/gb-2003-5-1-r1}
}

@article{hofacker1994fast,
  title={Fast Folding and Comparison of RNA Secondary Structures},
  author={Hofacker, Ivo L and Fontana, Walter and Stadler, Peter F and Bonhoeffer, L. S. and Tacker, Manfred and Schuster, Peter},
  journal={Monatshefte f{\"u}r Chemie},
  volume={125},
  number={2},
  pages={167--188},
  year={1994},
  doi={10.1007/BF00818163}
}

Copy citation

proto-bio/proto-tools/proto_tools/tools/gene_annotation/miranda

View source

Open Notebook

Open notebook

Function	Description
run_miranda_scan()	Predict microRNA target sites in RNA/DNA sequences using miRanda (complementarity alignment plus …	Docs Source

License: miRanda’s own code is licensed under GPL-2.0, and it federates over bundled data sources and components, each under its own license terms.

Bundled dependencies, each under its own license:

• ViennaRNA RNAlib (bundled, v1.5b): ViennaRNA Package License

Review each source’s terms before commercial use or redistribution.

Background

microRNAs are short (~22 nt) non-coding RNAs that regulate gene expression by base-pairing with complementary sites in target mRNAs — most often in the 3’ untranslated region (3’UTR) — and directing those transcripts toward translational repression or degradation. Because pairing is only partial in animals and concentrated in a short 5’ “seed” of the microRNA, identifying genuine target sites from sequence alone requires a method tuned to complementarity rather than ordinary sequence identity.

miRanda (Enright et al., 2003) addresses this with a two-phase algorithm. First, a Smith-Waterman local alignment scores microRNA-to-target complementarity (rewarding A:U and G:C pairs, and tolerating G:U wobble) instead of identity, with extra weighting on the microRNA’s 5’ seed region to reflect its outsized role in binding. Second, for alignments that clear a score threshold, the bundled ViennaRNA RNAlib (Hofacker et al., 1994) estimates the duplex minimum free energy (ΔG, kcal/mol). Sites that pass both the score and the energy threshold are reported as candidate target sites.

Learning Resources

• hacktrackgnulinux/miranda (maintained fork) - the source repository, with the canonical command-line flag surface and build instructions.
• miRBase (Griffiths-Jones et al.) - the reference registry of published microRNA sequences and names, and the standard place to obtain the microRNA queries you scan with.

Tools

miRanda Target Scan (miranda-scan)

Scans one or more microRNA queries against one or more RNA/DNA target sequences and returns, per target, the predicted target sites — each with its complementarity score, ViennaRNA duplex free energy, 1-indexed inclusive coordinates on both strands, percent identity and similarity, and the aligned strings.

API Reference

Source

Input: MirandaInput

target_sequences · List[string] · required

RNA/DNA target sequences (mRNA, 3’UTR, genomic) to scan.

mirna_queries · List[string] · required

microRNA query sequences, applied to every target.

mirna_ids · array

Optional microRNA labels (default seq_0, seq_1, …).

Source

Config: MirandaConfig

score_threshold · integer · default: "50"

Minimum alignment score to report a site (-sc).

energy_threshold · integer · default: "-20"

Maximum duplex free energy in kcal/mol; negative (-en).

scale · number · default: "4.0"

Contrast scaling on the microRNA 5’ seed region (-scale).

gap_open · integer · default: "-8"

Penalty for opening an alignment gap; negative (-go).

gap_extend · integer · default: "-2"

Per-position penalty for extending a gap; negative (-ge).

strict · boolean · default: "True"

Apply strict miRNA:target duplex base-pairing heuristics; -loose when False.

compute_energy · boolean · default: "True"

Compute ViennaRNA free energy; -noenergy when False.

trim · integer · default: "0"

Trim targets to this many nucleotides; 0 disables (-trim).

verbose · integer · default: "0"

Verbosity level (0=quiet, 1=info, 2=debug, 3=raw subprocess stderr). True is coerced to 1 and False to 0.

device · string · default: "cpu"

Device to run the tool on.

timeout · integer · default: "600"

Maximum execution time in seconds. None waits indefinitely.

seed · integer

Random seed. When set, tools run reproducibly up to small GPU float noise (see BaseToolOutput.approx_equal), and the seed participates in cache keys. When None, cacheable seed-sensitive tools skip cache until seeded.

Source

Output: MirandaOutput

results · List[MirandaSequenceResult]

Per-target results, in input order.

Show MirandaSequenceResult

target_id · string · required

Identifier of the target sequence

target_sequence · string · required

The input target sequence

target_sites · List[MirandaTargetSite]

Predicted microRNA target sites in this sequence, sorted by score descending

Applications

Use this to nominate candidate microRNA target sites in a transcript before committing to experimental validation. Typical workflows scan a small panel of microRNAs (from miRBase) against a gene’s 3’UTR to rank putative binding sites, or screen one microRNA across a set of candidate target transcripts to shortlist which are most likely regulated.

Usage Tips

• score_threshold and energy_threshold set the sensitivity-versus-specificity trade-off. Lowering score_threshold below the default of 50 and raising energy_threshold toward 0 from the default of -20 kcal/mol both admit weaker sites and recover more candidates at the cost of more false positives; tighten them in the other direction for high-confidence calls.
• Disable strict for more sensitivity. strict applies stringent miRNA:target duplex base-pairing heuristics by default; turning it off (-loose) is less conservative and surfaces additional, lower-confidence sites.
• Disable compute_energy for a much faster score-only scan. compute_energy runs the ViennaRNA free-energy phase by default; turning it off skips that phase entirely (the reported energy is then 0.0), which is useful for quick large-scale complementarity sweeps where ΔG ranking is not yet needed.
• All reported positions are 1-indexed and inclusive on both the target and the microRNA, matching biological coordinate conventions.

Toolkit Notes

These apply to every miRanda tool in this toolkit (miranda-scan).

• microRNA queries are user-supplied. miRanda does not ship a microRNA database; obtain query sequences from miRBase (or your own designs) and pass them as mirna_queries.
• Target sequences should be mRNAs or 3’UTRs, not whole genomes. The aligner builds a query-by-target dynamic-programming matrix, so memory scales with target length. Scan transcript-scale sequences, and use trim to cap very long targets rather than feeding whole chromosomes.
• Statistical shuffling and Z-scores are not supported in this fork. The randomization-based Z-score significance machinery is non-functional upstream and is not exposed by the wrapper; rank candidates by score and free energy instead.

Example notebook: See the full working example for a copy-paste-ready walkthrough.

Infrastructure Guides

The following guides cover how to run tools efficiently and at scale.

Tool Persistence

Keep a tool’s model warm across calls instead of reloading it every invocation.

Device Management

How GPUs are allocated to tools and how to target specific devices.

Parallel Execution

Fan a batch of inputs out across multiple GPUs.

Cloud Inference

Run tools on managed cloud infrastructure with no local setup.