Repository navigation
Documentation
K-tools links sequence enrichment to candidate post-transcriptional regulators. Select RNA populations from your experiment, count overlapping k-mers, extract a signature, and compare that signature with RBP-associated sequence preferences.
- Install the package and its dependencies.
- Read the parameter guide: exact thresholds, numerical examples, background construction and output interpretation.
- Run the simulated tutorial and compare the included demo output.
- Use the API reference for complete call signatures and methodology for the statistical definitions.
Use KEA when you have a reference group (for example, RNAs sharing an experimental property) and a meaningful control group. At k=7 there are 4⁷ = 16,384 sequence types. Two complementary analyses compare pooled group frequencies and per-transcript frequency distributions. Their intersection defines the integrated signature.
The code also exposes a rank-difference approach and a continuous score-correlation analysis. These are distinct selections, not interchangeable statistical significance estimates. Methodology describes their definitions.
KRS ranks the target's normalized frequency separately for each k-mer among transcripts containing that k-mer. A k-mer at percentile ≥0.95 enters the default signature. The target participates in the ranking. This is not a selection of the top 5% of k-mer types within the target.
K-RBP compares signature and background scores for each RBP–cell-line profile. A one-sided Mann–Whitney U test asks whether signature scores are higher; multiple-testing corrections and Cliff's delta support interpretation. When working from 7-mers against a 5-mer matrix, overlapping 5-mers are deduplicated before analysis.
RBP association depends on the reference matrix and experimental context. It is not a calibrated interaction probability, proof of direct binding, or evidence that other RBPs are absent.
| Input | Format | Needed for |
|---|---|---|
| Reference/control sequences | FASTA, uppercase DNA alphabet A/C/G/T, transcript orientation | KEA |
| Target plus background sequences | FASTA, unique identifiers; target included exactly once | KRS |
| RBP reference matrix | Numeric TSV, k-mers as rows, RBP profiles as columns | Biological K-RBP |
| Transcript scores | Two-column DataFrame: transcript identifier and numeric score | Correlation analysis |
| Transcript annotation or FASTA | Annotation with ID or ensembl_transcript_id and cdna, or FASTA |
Positional profiling |
FASTA identifiers are the first whitespace-delimited token after >. Convert U to T for counting, resolve ambiguous nucleotides, and keep identifiers disjoint across reference/control files. The current counter does not automatically standardize the alphabet. Non-ACGT windows do not match the enumerated k-mer universe. Use sequences at least k nucleotides long.
Use matching annotation releases for sequences, region boundaries and mapped experimental observations.
KEA writes all_kmers.tsv, enriched.txt and depleted.txt for each requested selection. KRS writes signature.txt and kmer_ranks.tsv; the full rank matrix is optional and can be large. K-RBP returns a DataFrame; save it explicitly with to_csv. The API reference records parameters, return values and output locations.
The supported tutorial covers FASTA counting, KEA, KRS and continuous-mode K-RBP. Advanced genomic mapping and BEDTools helpers require separately configured tools and resources. See Known limitations before using them.
- Installation
- Tutorial
- Methodology
- Detailed parameter guide
- API reference
- Data and reproducibility
- FAQs
- Citation
- K-map: similarity metrics, parameters and Sankey visualization
- K-CDS: inputs, frame, backgrounds and Fisher tests
Rank candidate transcripts against a query using complete normalized k-mer profiles. Visualize selected pairs against a background using binned Sankey-style rank alignment. See K-map.
Test selected CDS occurrence overlaps with UniProt genomic features using control transcripts or a shuffled-position null. Inputs, frame filtering, QC and statistical interpretation are described in K-CDS.
miRNA family annotation: orientation, parameters, scores and plots · CDR1as worked example