Documentation · Tutorial · Known limitations

Import with from KEA import KEA. The class exposes the analysis modules. Signatures below are extracted from the distributed code; scientific interpretation is described in Methodology.

For the behavior and practical effect of the parameters below, see the detailed parameter guide.

__init__(self, dir_out, ref_fasta, ctrl_fasta)

Creates the output directory and stores reference/control FASTA paths. It does not load or count sequences.

CreateCombinedFasta(self, species=None)

Concatenates input FASTA files and stores combined_fasta. Call before counting. Inputs must end with a newline and use distinct IDs. Returns None.

KmersCountsTable(self, k, cores=5, subset_by_subseq=None, species=None)

Enumerates A/C/G/T k-mers, counts with a multiprocessing pool, and stores the table in kmers_count_table_by_species["input"] for default inputs. The k column is metadata. Returns None. Use a guarded Python script.

ExtractKmers(self, k_selected, species=None, extraction_methods=('delta_rank_percentile', 'delta_median', 'stat_log2fc'), n_iter=1000, sign_emp_pval=0.0025, log2fc_threshold=1.0, fdr_alpha=0.05, stat_test='mwu', correction_method='bonferroni', median=True)

Returns KEA_results, nested by species key then method. Each method holds enriched, depleted and all_kmers. Saves tables in ExtractKmers/<species>/<method_parameters>/. Request the two integrated methods explicitly. The default tail fraction is 0.0025.

KRS(self, reference, species=None, top_pct=5, store=True, save=True, save_full_rank_table=False, verbose=True)

Returns a list of signature k-mers. Stores a KRS result dictionary and, by default, saves ExtractKmers/<species>/KRS/<safe_target_id>/signature.txt and kmer_ranks.tsv. top_pct=5 means rank ≥0.95.

KmersProteinEnrichment(self, kmers, df_zscore, k_out=None, deconvolute_unique=True, convert_to_rna=True, mode='continuous', test='mwu', threshold=None, top_n=None, fdr_alpha=0.05, plot=True, p_col='fdr', p_thresh=0.05, fc_thresh=1.0, highlight=None, highlight_top_n=10, save=None, bg_kmers=None, plot_type='lollipop', wordcloud_score='cliff_delta', wordcloud_top_n=10, wordcloud_font_path=None)

Returns a DataFrame containing all evaluated profiles. Core fields: protein, mean_group1, mean_group2, median_group1, median_group2, pvalue, fdr, bonf, zscore_diff (Cliff's delta), n1, n2. save saves figures, not the result table: call to_csv yourself. Use continuous mode and explicit plot thresholds. Historical binary-mode parameters are not implemented.

KmerCorrelationWithScore(self, txids_and_score: pd.DataFrame, species=None, method='spearman', sign_emp_pval=0.05)

Accepts a two-column DataFrame (transcript ID, numeric score). Returns a correlation table and writes KmerCorrelationWithScore/<species>/corr_analysis.<method>.txt plus tail lists. For default inputs pass species="input"; ensure KEA_results is a dictionary.

KmerSlidingWindowProfile(self, kmers, tx_id, ann_df, fasta_file=None, kmer_groups=None, window_size=50, step=25, strand='+', n_clusters='auto', cluster_range=(2, 10), convert_to_rna=True, plot_line=True, plot_heatmap=True, plot_cluster_heatmap=True, plot_silhouette=True, save_prefix=None, verbose=True, improvement_threshold=0.1)

Accepts transcript sequence via fasta_file or annotation (ID/ensembl_transcript_id and cdna). Counts and clusters positional signatures. Advanced functionality not exercised by the minimal tutorial.

Genomic interval and mapping helpers require external tools and configured resources.

• get_kmer_table_for_species
• LogoByList
• CompareTranscriptKmerProfiles
• plot_kmer_rank_alignment
• FetchOrthologsAndSaveFastaOneToOne
• IntersectEnrichedKmersAndPlotVenn
• ImportMetaTx
• ImportMetaTx_Mapped
• ImportMetaTx_MappedScore
• ImportMetaTx_Enrichment
• GenerateMetaTx
• MapperTxToGenome
• ScoreByBigWig
• EnrichmentAnalysis
• ScoreMatrix_Enrichment
• KmerBinFrequencyMatrix
• PlotKmerMatrixFreq

kmers_counter(seq, k, all_kmers_df=None, tx_id=None) counts overlapping windows using the reference-class boundary, which omits the final possible start. If an enumerated k-mer universe is supplied, absent entries are filled with zero.

run_stat_test(x, y, test) supports mwu, ks and ttest; it skips empty or jointly constant groups. Other helper functions support interval handling, annotations and plotting and are outside the portable workflow.

• K-map: similarity metrics, parameters and Sankey visualization
• K-CDS: inputs, frame, backgrounds and Fisher tests

K_CDS(self, gtf_file=None, uniprot_dir=None,
    kmers=None, method="stat_log2fc", species=None,
    in_frame=False, background="control", tracks=("unipDomain",),
    alpha=0.05, fdr_scope="global", split_blocks=True,
    seed=42, output_dir=None, plot=True, max_domains=60,
    ref_fasta=None, ctrl_fasta=None)

Returns a dictionary of statistics, occurrences, QC, matrices, figure, settings and output directory; also stores self.domain_enrichment. Supply GTF and UniProt paths explicitly. Optional FASTA overrides apply only to this call. See complete parameter explanations.

CompareTranscriptKmerProfiles(self, target_tx, other_txs,
    target_species=None, other_species=None, metric="spearman",
    pseudocount=1e-9, k=None, fasta_input=False,
    fasta_species_key="_external_db")

plot_kmer_rank_alignment(self, ref_tx, ctrl_tx,
    ref_species="input", ctrl_species="input", bg_species="input",
    kmers=None, exclude_ctrl_from_bg=True, top_n=None,
    min_freq=None, bin_size=100, alpha_band=0.4, color_by="gc",
    kmer_categories=None, figsize=(10, 10), save=None)

See K-map for metric definitions, cached external FASTAs, ranked output and Sankey-style visualization.

AnnotateMiRNASeeds(self, kmers=None, source="KEA", reference=None,
    species=None, species_id=9606, plot=True, top_n=20, output_dir=None)

analysis.seed_family_file selects the family table. The default is the resource installed with K-tools.

miRNA family annotation: orientation, parameters, scores and plots · CDR1as worked example

K-CDS uses K_CDS as its Python method name; The former DomainEnrichment name is no longer exposed. Results also include frame_distribution and frame_figure, and are stored as self.k_cds. See frame plot and saved outputs.

With plot=True, the domain heatmap and CODING/SHUFFLE frame distribution are separate figures. Display result["figure"] for the heatmap and result["frame_figure"] for the frame plot. The frame legend is outside the SHUFFLE panel. Figures are saved as heatmap.png/.pdf and frame_distribution.png/.pdf.