Bio Splice Variant Prediction

Version Compatibility

Reference examples tested with: SpliceAI 1.3+, Pangolin 1.0+, MMSplice 2.4+, pyensembl 2.3+, pysam 0.22+, pandas 2.2+, gffutils 0.13+, tensorflow 2.15+

Before using code patterns, verify installed versions match. If versions differ:

• Python: pip show <package> then help(module.function) to check signatures
• CLI: <tool> --version then <tool> --help to confirm flags

If code throws ImportError, AttributeError, or TypeError, introspect the installed package and adapt the example to match the actual API rather than retrying.

Splice Variant Prediction

Predict whether a DNA variant alters mRNA splicing. Distinct from "variant pathogenicity" generally: a variant can be a strong splice disruptor without being pathogenic for the gene's standard mechanism, or pathogenic for reasons orthogonal to splicing. Splice prediction asks specifically: does this variant change splice-site usage?

Predictor Taxonomy

Tool Selection Matrix

Methodology evolves; verify benchmarks (Strawn 2025 bioRxiv; You et al 2024 Nat Commun) and ClinGen SVI splicing recommendations before reporting clinical interpretations. Concordance across SpliceAI + Pangolin + MMSplice is gold-standard evidence; discordance flags need RNA validation.

Decision Tree by Use Case

ClinGen SVI 2023 Framework

The ClinGen Sequence Variant Interpretation (SVI) splicing subgroup (Walker 2023 Am J Hum Genet) extended the ACMG/AMP 2015 framework with explicit splice-prediction rules.

Operational rules: Computational evidence (PP3/BP4) is supporting, not standalone. Splicing variants benefit from concordance across SpliceAI + Pangolin + MMSplice. RNA validation supersedes prediction. Always log SpliceAI version, distance window, and reference transcript. SpliceAI alone is not sufficient for PVS1; canonical site disruption requires gene-level LoF context.

SpliceAI Workflow

Goal: Annotate VCF variants with per-variant delta scores for splice-site change.

Approach: Run spliceai CLI with reference genome and annotation; parse INFO field for delta scores. SpliceAI is human-only (-A grch37 or -A grch38); the model was trained on GENCODE human and does not directly transfer to mouse, fly, or other species. For mouse, retrained variants exist (e.g. mouseSpliceAI); for other species, use Pangolin (4 species: human, mouse, rat, rhesus macaque) or accept that prediction will be unreliable.

spliceai \
    -I input.vcf \
    -O output.vcf \
    -R GRCh38.primary_assembly.genome.fa \
    -A grch38 \
    -D 50 \
    -M 0

-D 50 = distance window in nt around variant (default 50). For deep-intronic variants suspected of creating pseudoexons, raise to 500-2000:

spliceai -I input.vcf -O output_extended.vcf -R genome.fa -A grch38 -D 500 -M 1

-M 0 (default) returns raw scores; -M 1 masks splice gains at annotated sites and losses at unannotated sites (cleaner for clinical use). Output INFO format: SpliceAI=ALLELE|SYMBOL|DS_AG|DS_AL|DS_DG|DS_DL|DP_AG|DP_AL|DP_DG|DP_DL. Delta score = max(DS_AG, DS_AL, DS_DG, DS_DL).

import pandas as pd
import re

def parse_spliceai_vcf(vcf_path):
    rows = []
    with open(vcf_path) as f:
        for line in f:
            if line.startswith('#'):
                continue
            fields = line.strip().split('\t')
            info = fields[7]
            m = re.search(r'SpliceAI=([^;]+)', info)
            if not m:
                continue
            for ann in m.group(1).split(','):
                parts = ann.split('|')
                allele, symbol = parts[0], parts[1]
                ds = [float(p) if p != '.' else 0 for p in parts[2:6]]
                dp = parts[6:10]
                rows.append({
                    'chrom': fields[0], 'pos': int(fields[1]),
                    'ref': fields[3], 'alt': allele,
                    'gene': symbol,
                    'DS_AG': ds[0], 'DS_AL': ds[1],
                    'DS_DG': ds[2], 'DS_DL': ds[3],
                    'delta_max': max(ds),
                })
    return pd.DataFrame(rows)

df = parse_spliceai_vcf('output.vcf')
df['acmg_evidence'] = pd.cut(
    df['delta_max'],
    bins=[-0.01, 0.10, 0.20, 0.50, 0.80, 1.01],
    labels=['BP4', 'inconclusive', 'PP3_supporting', 'PP3_moderate', 'PP3_strong']
)

DS labels: AG = acceptor gain, AL = acceptor loss, DG = donor gain, DL = donor loss.

Pangolin for Tissue-Specific Prediction

Goal: Get tissue-specific splice impact predictions when disease tissue is known.

Approach: Run Pangolin CLI with VCF + reference + gffutils annotation database.

python -c "import gffutils; gffutils.create_db('gencode.v45.annotation.gff3', 'gencode.db', force=True)"

pangolin \
    input.vcf \
    GRCh38.primary_assembly.genome.fa \
    gencode.db \
    pangolin_output \
    -d 500 \
    -m True \
    -s 0.2

-m True masks splice gains at annotated sites and losses at unannotated sites (recommended for clinical use). -s 0.2 outputs all sites with predicted change >= cutoff.

Pangolin output is a VCF with per-tissue predictions across the 4 tissues used at training: brain, heart, liver, testis (Zeng & Li 2022 Genome Biol). The model outputs per-species per-tissue predictions but extrapolates poorly to tissues outside this set. Use the tissue closest to disease-relevant context. For tissues not in the 4-tissue training set, fall back to SpliceAI — Pangolin extrapolates poorly to unseen tissues.

SpliceVault for Empirical Mis-Splicing Outcomes

Goal: Predict the type of mis-splicing (exon skipping vs cryptic site activation) given a canonical-disrupting variant.

Approach: Query SpliceVault's database of empirical mis-splicing events from public RNA-seq.

import requests

# Web API: https://kidsneuro.shinyapps.io/splicevault/
# Or use the R/Python package at github.com/kidsneuro-lab/SpliceVault

# Example: NM_000546.6:c.673-2A>G (TP53)
# Returns top-N most likely mis-splicing events: exon skipping, cryptic 3'ss usage, etc.

SpliceVault (Dawes 2023 Nat Genet) showed that the Top-4 events at any splice site explain >95% of empirical mis-splicing — a striking regularity that makes consequence prediction tractable. Use SpliceVault when the question is not "will splicing change?" but "what specific aberrant splicing will occur?".

MMSplice for Calibrated ΔPSI

Goal: Predict quantitative ΔPSI (not just probability of disruption) for cassette exons.

Approach: Score variant impact on each splicing region (5'ss, 3'ss, exon, intron-3'/5') and combine.

from mmsplice.vcf_dataloader import SplicingVCFDataloader
from mmsplice import MMSplice, predict_save

dl = SplicingVCFDataloader(
    gtf='gencode.v45.basic.gtf',
    fasta_file='GRCh38.fa',
    vcf_file='input.vcf'
)

model = MMSplice()
predict_save(model, dl, 'mmsplice_predictions.csv', pathogenicity=True)

MMSplice (Cheng 2019 Genome Biol) reports Δlogit_psi per variant. Useful when calibrated effect sizes matter (research) more than probability of disruption (clinical screening). Companion MTSplice (Cheng 2021 Genome Biol) adds tissue-specific Δψ predictions.

HGVS Splicing Nomenclature

Following den Dunnen 2016 Hum Mutat:

Validation tools: VariantValidator (Freeman 2018 Hum Mutat), Mutalyzer 3 (Lefter 2021 Hum Mutat).

Extended-Window Scoring for Deep-Intronic Variants

SpliceAI's default precomputed scores use a 50-nt window, missing variants that create pseudoexons in deep intronic regions. For unsolved Mendelian cases:

# Recompute with extended window
spliceai -I input.vcf -O output_2kb.vcf -R genome.fa -A grch38 -D 2000

# Or use CI-SpliceAI (Strauch 2022 Bioinformatics) optimized for distal effects

Pseudoexon creation in deep introns explains a substantial fraction of unsolved Mendelian disease alleles in current cohorts (estimates 5-15% across studies; specific quantitative range will vary by cohort and panel — verify against current literature). Disease examples: CFTR 3849+10kbC>T, USH2A c.7595-2144A>G, CEP290 c.2991+1655A>G (LCA10).

Concordance Across Predictors

import pandas as pd

merged = (spliceai_df
    .merge(pangolin_df, on=['chrom', 'pos', 'alt'], suffixes=('_sai', '_pang'))
    .merge(mmsplice_df, on=['chrom', 'pos', 'alt'])
)

merged['concordance'] = (
    (merged['delta_max_sai'] >= 0.2).astype(int) +
    (merged['pangolin_score'].abs() >= 0.2).astype(int) +
    (merged['delta_logit_psi'].abs() >= 1.0).astype(int)
)

merged['interpretation'] = merged['concordance'].map({
    0: 'concordant_benign',
    1: 'discordant_low_evidence',
    2: 'concordant_evidence',
    3: 'high_concordance_pathogenic'
})

Discordance is the most informative pattern — variants where one model sees impact and others don't are high priority for RNA validation.

Branchpoint Variant Detection

All current tools are weak at branchpoint variants because the BPS motif (yUnAy) has low information content. Specific branchpoint tools:

Branchpoint variants are under-recognized in clinical pipelines; SpliceAI captures only some because branchpoint motifs have low information content. Recommendation: when SpliceAI delta is borderline (0.1-0.3) for a variant in the BPS region (-18 to -40 from 3'ss), run BPHunter as supplement.

Splice-Switching ASO Design

Goal: Design antisense oligonucleotides to modulate splicing therapeutically (e.g. SMA ISS-N1, DMD exon skipping).

Approach: Use SpliceAI to predict impact of binding-site occlusion; check accessibility (RNAfold); avoid SR/hnRNP off-target motifs.

# Conceptual workflow - actual design uses ASO synthesis platforms
# 1. Identify target ESE/ESS/ISE/ISS region from MaxEntScan + SpliceAI scan
# 2. Design candidate 18-22 nt ASOs spanning the regulatory element
# 3. For each ASO, simulate splice-site occlusion impact via SpliceAI on the masked sequence
# 4. Filter for RNA accessibility (avoid stable hairpins) using RNAfold
# 5. Whole-transcriptome SpliceAI scan for off-target binding (>=17/20 nt match)
# 6. Avoid TLR9 immunostimulatory CpG motifs

# Chemistry choices:
# - 2'-MOE-PS: nusinersen-like (CNS, intrathecal)
# - PMO: DMD ASOs (systemic IV)
# - GalNAc-conjugated: hepatic targeting

Approved precedents: nusinersen (SMA ISS-N1 occlusion, exon 7 inclusion); risdiplam (small-molecule SMN2 splicing modulator); eteplirsen/golodirsen/casimersen/viltolarsen (DMD exon skipping). Design references: Hua 2008 AJHG; Aartsma-Rus 2023 Nat Rev Drug Discov.

Per-Tool Failure Modes

SpliceAI: 50nt Window Limitation

Trigger: Variant deep in an intron (>50 nt from canonical splice site).

Mechanism: Default precomputed scores use ±50 nt window; the model is trained on this context but pre-stored scores limit lookups.

Symptom: Known pathogenic deep-intronic variant scores low (<0.2); no pseudoexon detected.

Fix: Re-run with -D 500 or -D 2000; or use CI-SpliceAI optimized for distal effects.

SpliceAI: Tissue Agnosticism

Trigger: Variant in a tissue-specific gene (NEFM in neurons, MAPT brain, DMD muscle isoforms).

Mechanism: SpliceAI is trained on aggregate GENCODE annotation; tissue-specific events with weak constitutive use score low.

Symptom: Tissue-specific pathogenic variant has low SpliceAI delta; functional impact still observed in target tissue.

Fix: Use Pangolin for tissue-aware prediction; or SpliceTransformer; require RNA validation in disease-relevant tissue.

Pangolin: Out-of-Training Tissue

Trigger: Disease tissue not represented in Pangolin's 4-species, GTEx-tissue training set.

Mechanism: Pangolin extrapolates poorly to tissues outside training distribution.

Symptom: Pangolin score uncalibrated for queried tissue; doesn't agree with patient RNA-seq from that tissue.

Fix: Fall back to SpliceAI for tissues not in Pangolin training; or run patient RNA-seq directly.

MMSplice: Atypical Events

Trigger: Variant affecting a non-cassette event (MXE, complex multi-junction, AFE/ALE).

Mechanism: MMSplice modular model is trained primarily on cassette exon events.

Symptom: MMSplice ΔPSI doesn't match other predictors or empirical data for non-cassette events.

Fix: Use SpliceAI for non-cassette events; restrict MMSplice to cassette exon contexts.

CADD-Splice: Loss of Component Information

Trigger: Wanting to know which sub-component drove a high CADD-Splice score.

Mechanism: CADD-Splice combines SpliceAI + MMSplice + CADD into a single C-score; sub-component contributions are abstracted.

Symptom: "High CADD-Splice score but unclear why."

Fix: Run SpliceAI and MMSplice separately to see which contributed.

Branchpoint Variants: Low Information Motif

Trigger: Variant in the BPS region (-18 to -40 from 3'ss).

Mechanism: BPS motif (yUnAy) has low information content; CNNs struggle to learn the consensus.

Symptom: Confirmed BPS variant scores SpliceAI delta <0.2 despite functional disruption.

Fix: Use BPHunter (Zhang 2022 PNAS) for genome-wide branchpoint screening; require RNA validation.

Population Database Lookup

Always check ClinVar first for existing classifications; cross-reference with gnomAD for population frequency before committing to PP3/PP4.

Common Errors

Common Pitfalls

• Using SpliceAI score alone for clinical reporting — must combine with concordant predictors and ideally RNA validation; ClinGen SVI requires this for non-canonical positions.
• 50nt window for deep intronic variants — pseudoexon-creating variants 100-2000 nt deep are systematically missed.
• Tissue-agnostic prediction for tissue-specific genes — use Pangolin or SpliceTransformer when tissue context matters (NEFM, MAPT, DMD isoforms).
• Branchpoint variants — all current predictors are weak here. Use BPHunter for branchpoint screening.
• Forgetting NMD direction — confirmed splice disruption needs NMD-status check. Last-exon PTCs escape NMD and can be dominant-negative or gain-of-function.
• In-silico-only PVS1 application — PVS1 for non-canonical positions requires functional or strong computational evidence; SpliceAI alone is supporting (PP3), not very strong.
• Trusting LLMs for variant interpretation — use as orchestrators on top of SpliceAI/VariantValidator/ClinVar; all clinical-grade calls require human expert sign-off.
• Skipping HGVS validation — invalid HGVS leads to silent reference-transcript mismatches; always run VariantValidator first.

Quality Thresholds

Related Skills

• splicing-qc - MaxEntScan + library QC for confirming predicted impact
• splicing-quantification - Empirical PSI from RNA-seq to validate predictions
• outlier-splicing-detection - FRASER2/DROP for RNA-seq confirmation in clinical samples
• variant-calling/clinical-interpretation - Broader ACMG/AMP variant interpretation framework
• variant-calling/variant-annotation - VEP plugin integration for SpliceAI

References

• Jaganathan et al 2019 Cell - SpliceAI
• Zeng & Li 2022 Genome Biol - Pangolin
• Cheng et al 2019 Genome Biol - MMSplice
• Cheng et al 2021 Genome Biol - MTSplice (tissue MMSplice)
• You et al 2024 Nat Commun - SpliceTransformer
• Strauch et al 2022 Bioinformatics - CI-SpliceAI extended window
• Rentzsch et al 2021 Genome Med - CADD-Splice
• Dawes et al 2023 Nat Genet - SpliceVault
• Walker et al 2023 Am J Hum Genet - ClinGen SVI splicing recommendations
• Riepe et al 2021 Hum Mutat 42:799 - SpliceAI in clinical pipelines (Riepe TV et al)
• Abou Tayoun et al 2018 Hum Mutat - PVS1 decision tree
• Richards et al 2015 Genet Med - ACMG/AMP framework
• den Dunnen et al 2016 Hum Mutat - HGVS standard
• Zhang et al 2022 PNAS (PMID 36306325) - BPHunter for branchpoints
• Hua et al 2008 AJHG - ISS-N1 / nusinersen mechanism
• Aartsma-Rus 2023 Nat Rev Drug Discov - DMD exon-skipping ASOs
• (Extended-window SpliceAI in unsolved Mendelian: consult current literature for citation — earlier draft attribution to "Smith 2024 Nat Commun" could not be verified.)
• Findlay et al 2018 Nature - BRCA1 saturation genome editing (MAVE)