FUNGUS-SV: Interactive Structural Variant Report — CICC-1445 vs 5 Reference Strains

Title: FUNGUS-SV: Structural Variant Discovery and Depth-Centric Validation for Haploid Fungal Genomes v1.1 Status: Active development Observations:FUNGUS-SV is a hypothesis-generation tool intended to save time when screening for structural variants (SVs) in haploid fungal genomes, using empirically calibrated, real-data-derived parameters. It is not intended as a substitute for orthogonal experimental validation of any individual candidate SV. Authors:Guimarães, Kelton H. A. Description:This dataset contains an interactive HTML report of structural variants (SVs) detected between the query strain CICC-1445 and five reference strains of Saccharomyces cerevisiae (S288C, BJ4, IMX2600, Makgeolli, SX2), scored using a depth-centric confidence pipeline calibrated for haploid genomes. Methods:PacBio HiFi reads from CICC-1445 (SRR18210299, 274,915 reads, ~20 kb N50) were aligned with minimap2 (map-hifi preset). SVs were called with Sniffles2, cuteSV, and SVIM, and merged via an Iterative Consensus Builder (ICB) requiring agreement from at least two of three callers (≥50% reciprocal overlap, 200 bp flank tolerance). Candidate SVs are scored using a depth-centric confidence engine: depth signature (DHFFC + DHBFC, weight 0.65, the primary discriminative layer), Local Assembly Refinement (Flye + Miniasm/Racon, weight 0.20, run on selected candidates), and k-mer spectrum (implemented, not used in this release). Breakpoint-junction split-read support is computed and reported for every candidate as a confirmatory annotation. Duplications and inversions with strong breakpoint-junction support are additionally flagged via a type-specific override, since depth signal is known to fail to detect these variant types in haploid genomes. Reference genomes: S288C (NCBI R64-3-1), BJ4 (ENA), IMX2600 (GenBank), Makgeolli (GenBank), SX2 (ENA). Gene annotation from NCBI RefSeq GFF (6,459 genes). Results Summary — Total SVs (ICB consensus): Reference Relationship to CICC-1445 Total SVs S288C Within-species 285 BJ4 Within-species 176 IMX2600 Within-species 345 Makgeolli Within-species 244 SX2 Within-species 310 S. jurei Cross-species (baseline, not directly comparable) 167 Confidence Tier Results: Strain TRIPLE (T≥0.80) DOUBLE (0.60–0.79) SINGLE (0.40–0.59) WEAK (0.20–0.39) CONTRADICTED (<0.20) DUP override INV override S288C 49 (17.2%) 68 13 26 98 19 12 BJ4 19 (10.8%) 20 9 22 81 14 11 IMX2600 40 (11.6%) 80 28 28 137 20 12 Makgeolli 21 (8.6%) 56 22 26 93 17 9 SX2 17 (5.5%) 83 16 29 138 18 9 Jurei (cross-species) 12 40 3 5 96 11 0 Note on S. jurei: S. jurei is a different species from the query strain (S. cerevisiae CICC-1445) and is included here as an informative cross-species baseline, not as a sixth within-species validation comparison. Sequence divergence between species causes reads to map incorrectly to the reference, inflating apparent candidate SV counts; two-assembler Local Assembly Refinement confirmed 0 of 10 tested candidates from this comparison (0%), consistent with these calls being predominantly cross-species mapping artifacts rather than true structural variants. This result supports interpreting the pipeline's within-species results (above) as specific to genuine intra-species structural variation, rather than reflecting non-specific noise. Validation Highlights: Within-species confirmation via two-assembler Local Assembly Refinement (both assemblers required to agree): 51.7% raw, 69.6% excluding technical LAR failures (assembler timeout or assembly failure). Self-alignment negative controls (a genome aligned to itself, expected to yield zero true SVs) establish an empirical false-discovery rate of 0.8% and 0.0% at the TRIPLE_TRIANGULATED tier, across two fungal genera tested. Duplication and inversion calls rely on a split-read override correcting for depth's known blind spot for these variant types in haploid genomes, and show lower, more variable confirmation than deletion calls. Users should treat DUP/INV calls, particularly small duplications (≤2 kb), with additional caution. Larger SVs are detected more reliably than small SVs (<100 bp) across all evidence layers. Key Findings: CICC-1445 is closest to BJ4 (176 SVs), consistent with both being industrial strains. CICC-1445 is most divergent from IMX2600 (345 SVs), an engineered laboratory strain. FLO adhesin genes and Ty retrotransposons are recurrent hotspots of structural variation in this comparison. Source code: https://github.com/keltonjenkovguimaraes-alt/fungus-sv HTML results link: saccharomycesreport.netlify.appLicense: MIT