c05b0aa800df56a3fd1a5c5cbb903e39a54b6396
lrnassar
Fri Aug 14 10:42:09 2026 -0700
TP53 Provisional: nt-aware splicing, caveat 4, and per-nt divergence flags. refs #37399
From the ClinGen TP53 VCEP review. The Provisional summary is one row per missense
protein change, but splicing and allele frequency are nt-specific:
- Splicing PP3 (SpliceAI >= 0.2, supersedes a missense BP4) now applies only when
UNAMBIGUOUS: every nt path yielding the protein change is >= 0.2. When paths
straddle the threshold the row keeps its missense code (fixes e.g. L35F, where
c.105G>T=0.35 but c.105G>C=0.13 was wrongly flagged via the codon max).
- Caveat (GN009): when splicing PP3 is applied, drop the protein-level functional
assay code (PS3/BS3) from the point sum, since the assay tests the missense
protein not the splicing effect.
- New ntVerify column (bed9+10 -> 9+11, filterable): flags rows whose nt variants
disagree on SpliceAI or on gnomAD AF, with a mouseover pointer to verify the
exact nt change in the Bioinformatic / Allele Frequencies track.
load_s2 now tracks spliceai_min; af_code_and_points returns af_divergent (AF
aggregation unchanged). Reviewed clean by a fresh-eyes pass.
diff --git src/hg/makeDb/scripts/tp53/tp53ProvisionalClass.py src/hg/makeDb/scripts/tp53/tp53ProvisionalClass.py
index e50911ca9c7..fe3c1f6c28d 100644
--- src/hg/makeDb/scripts/tp53/tp53ProvisionalClass.py
+++ src/hg/makeDb/scripts/tp53/tp53ProvisionalClass.py
@@ -13,31 +13,31 @@
* Splicing PP3 (SpliceAI >= 0.2)
The point sum is bucketed into P / LP / VUS / LB / B per the CSpec
classification ranges. BA1 is stand-alone Benign and forces class = Benign
regardless of other evidence.
DELIBERATELY EXCLUDED from the sum (documented in every mouseover):
- PVS1 (null variants only; handled in separate track)
- PS1 / PS2 / PS4 / PP1 / PP4 / BS4 (require clinical observations)
- BP7 (computational, but synonymous/intronic only; out of scope for this
missense-only track)
This is NOT a ClinGen classification — the warning is in every mouseover
since clinicians live in the mouseover, not the description page.
-bigBed 9+10.
+bigBed 9+11.
"""
import argparse
import json
import os
import re
import sys
import openpyxl
sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
import tp53FuncLib as lib
DEFAULT_OUTDIR = "/hive/users/lrnassar/claude/RM37399/provisionalClass"
SRC_S3 = "/hive/users/lrnassar/claude/RM37399/tp53_downloads/Functional-worksheet.xlsx"
@@ -104,30 +104,31 @@
string name; "Missense change (e.g., R175H)"
uint score; "Not used, all 0"
char[1] strand; "Not used, all ."
uint thickStart; "Same as chromStart"
uint thickEnd; "Same as chromEnd"
uint reserved; "RGB color"
string provisionalClass; "Provisional class (Pathogenic/LP/VUS/LB/Benign)"
int totalPoints; "Sum of applied Tavtigian points"
string appliedCodes; "Codes contributing (semicolon-separated)"
string pm1; "PM1 contribution"
string ps3bs3; "PS3/BS3 contribution (from Table S3)"
string pp3bp4; "PP3/BP4 contribution (Table S2 + splicing)"
string af; "AF code (BA1/BS1/PM2_Supporting/none)"
string bs2; "BS2 evidence (FLOSSIES cohort observation)"
string spliceAI; "Max SpliceAI delta (from Table S2)"
+ string ntVerify; "nt-divergence flag: SpliceAI / gnomAD AF path-dependent, verify per-nt"
lstring _mouseOver; "HTML mouseover"
)
"""
def parse_missense_short(p):
if not isinstance(p, str):
return None
m = PROT_RE.match(p.strip())
if not m:
return None
return (m.group(1), int(m.group(2)), m.group(3))
def parse_hgvsp3(p):
@@ -148,64 +149,68 @@
ws = wb["Supplementary Table S3"]
out = {}
for row in ws.iter_rows(min_row=4, values_only=True):
k = parse_missense_short(row[0])
if not k:
continue
out[k] = {
'code': str(row[7]).strip() if row[7] is not None else 'No evidence',
}
return out
def load_s2(path):
"""Table S2: per c.X>Y missense with preliminary PP3/BP4 + max SpliceAI.
- Returns {(wt, codon, alt) -> {code, spliceai_max, paths: [(c_pos, ref_nt, alt_nt), ...]}}.
+ Returns {(wt, codon, alt) -> {code, spliceai (max), spliceai_min,
+ paths: [(c_pos, ref_nt, alt_nt), ...]}}.
'paths' enumerates every c.X>Y combination that yields the protein change —
used downstream to look up AF at each contributing genomic coord.
"""
wb = openpyxl.load_workbook(path, data_only=True)
ws = wb["Supplementary Table S2"]
out = {}
for row in ws.iter_rows(min_row=4, values_only=True):
k = parse_hgvsp3(row[1])
if not k:
continue
hgvsc = row[0]
path_tuple = None
if isinstance(hgvsc, str):
mc = HGVSC_RE.match(hgvsc.strip())
if mc:
path_tuple = (int(mc.group(1)), mc.group(2), mc.group(3))
new_code = str(row[4]).strip() if row[4] is not None else 'No evidence'
try:
new_spliceai = float(row[5]) if row[5] is not None else 0.0
except (ValueError, TypeError):
new_spliceai = 0.0
existing = out.get(k)
if existing is None:
out[k] = {
'code': new_code,
- 'spliceai': new_spliceai,
+ 'spliceai': new_spliceai, # max across nt paths
+ 'spliceai_min': new_spliceai, # min across nt paths (straddle test)
'paths': [path_tuple] if path_tuple else [],
}
else:
if _s2_priority(new_code) > _s2_priority(existing['code']):
existing['code'] = new_code
if new_spliceai > existing['spliceai']:
existing['spliceai'] = new_spliceai
+ if new_spliceai < existing['spliceai_min']:
+ existing['spliceai_min'] = new_spliceai
if path_tuple and path_tuple not in existing['paths']:
existing['paths'].append(path_tuple)
return out
def _s2_priority(code):
order = {
'PP3_moderate': 5, 'BP4_moderate': 5,
'PP3': 4, 'BP4': 4,
'No evidence': 1,
}
return order.get(code, 0)
def load_hotspot_occurrences(path):
@@ -338,66 +343,79 @@
return ('BP4_Moderate', -2)
return ('', 0)
def af_code_and_points(wt, codon, alt, paths, af_lookup, present_set, tx):
"""For a (wt, codon, alt) protein change, look up gnomAD v4.1 AF at every
c.X>Y path and return the strongest applicable code.
Priority (most-benign first): BA1 > BS1 > PM2_Supporting > none.
BA1 is stand-alone Benign; the caller forces classification = Benign.
PM2_Supporting applies either when a path is coded PM2 (present but rare) or
when every path is absent from gnomAD (present_set). A path that is present
in gnomAD but not rare enough to be coded blocks the absent-based PM2, so a
variant seen in gnomAD at moderate frequency is not given PM2.
+
+ Also returns af_divergent: True when the nt paths yield different AF codes
+ (e.g. one path absent from gnomAD -> PM2 while another is present at moderate
+ frequency -> no code). The protein-level row can only show one, so these are
+ flagged for per-nt verification in the AF track.
"""
chrom = tx['chrom']
strand = tx['strand']
rcomp = {'A':'T','T':'A','C':'G','G':'C'}
- seen = []
- any_present = False
- any_absent = False
+ per_path = [] # per nt path: 'BA1' / 'BS1' / 'PM2_Supporting' / 'absent' / 'other'
for c_pos, c_ref, c_alt in paths:
g = lib.cdna_coding_to_genomic(c_pos, tx)
if g is None:
continue
if strand == '-':
g_ref = rcomp.get(c_ref, c_ref)
g_alt = rcomp.get(c_alt, c_alt)
else:
g_ref = c_ref
g_alt = c_alt
# AF lookup uses 1-based start
key = (chrom, g + 1, g_ref, g_alt)
code = af_lookup.get(key)
if code:
- seen.append(code)
- if key in present_set:
- any_present = True
+ per_path.append(code)
+ elif key not in present_set:
+ per_path.append('absent') # absent from gnomAD -> PM2-eligible
else:
- any_absent = True
- if 'BA1' in seen:
- return ('BA1', 0, 'stand-alone Benign')
- if 'BS1' in seen:
- return ('BS1', POINTS['BS1'], '-4 pts')
- if 'PM2_Supporting' in seen:
- return ('PM2_Supporting', POINTS['PM2_Supporting'], '+1 pt')
+ per_path.append('other') # present but not rare enough -> no code
+ coded = [c for c in per_path if c not in ('absent', 'other')]
+ any_absent = 'absent' in per_path
+ any_present = ('other' in per_path) or bool(coded)
+
+ # Effective code per path (absent -> PM2, present-Other -> none); if the set
+ # disagrees the protein change is path-dependent for allele frequency.
+ eff = set('PM2_Supporting' if s == 'absent' else ('' if s == 'other' else s)
+ for s in per_path)
+ af_divergent = len(eff) > 1
+
+ if 'BA1' in coded:
+ return ('BA1', 0, 'stand-alone Benign', af_divergent)
+ if 'BS1' in coded:
+ return ('BS1', POINTS['BS1'], '-4 pts', af_divergent)
+ if 'PM2_Supporting' in coded:
+ return ('PM2_Supporting', POINTS['PM2_Supporting'], '+1 pt', af_divergent)
if any_absent and not any_present:
return ('PM2_Supporting', POINTS['PM2_Supporting'],
- '+1 pt (absent from gnomAD)')
- return ('', 0, '')
+ '+1 pt (absent from gnomAD)', af_divergent)
+ return ('', 0, '', af_divergent)
def bs2_evidence(wt, codon, paths, flossies_lookup, tx):
"""Return (bs2_label, points) for the strongest FLOSSIES BS2 tier matching
any (c.X>Y) path by exact genomic position AND ref/alt, else (None, 0).
BS2 requires the SAME nt change — an observation of c.1120G>A does
not support BS2 for c.1120G>C even though both yield p.G374R."""
chrom = tx['chrom']
strand = tx['strand']
rcomp = {'A':'T','T':'A','C':'G','G':'C'}
best = (None, 0)
for c_pos, c_ref, c_alt in paths:
g = lib.cdna_coding_to_genomic(c_pos, tx)
if g is None:
continue
@@ -416,204 +434,256 @@
CAVEATS_STR = (
"PVS1 (null variants only — see PVS1 tracks); "
"PS1/PS2/PS4/PP1/PP4/BS4 (require clinical observations); "
"BP7 (synonymous/intronic; N/A to missense)."
)
HEADER_WARNING = (
""
"NON-FINAL: NOT a ClinGen classification — preliminary point-sum only."
""
)
def mouseover(name, cls, pts, pm1, ps3, pp3, af_lbl, bs2_lbl, applied,
- spliceai, splice_pp3_active, splice_overrode_bp4, codon, ba1):
+ spliceai, splice_pp3_active, splice_overrode_bp4, codon, ba1,
+ spliceai_min=0.0, splice_straddle=False, af_divergent=False,
+ ps3_suppressed=''):
splice_section = ""
- if spliceai and spliceai > 0:
+ if splice_straddle:
+ splice_section = (
+ "
SpliceAI: {mn:.2f}–{mx:.2f} — "
+ "path-dependent: the nt variants at this "
+ "codon straddle the 0.2 threshold, so splicing PP3 is not applied at "
+ "the protein level. Verify the exact nt change in the Bioinformatic "
+ "track."
+ ).format(mn=spliceai_min, mx=spliceai)
+ elif spliceai and spliceai > 0:
if splice_pp3_active and splice_overrode_bp4:
splice_section = (
"
SpliceAI: {sa:.2f} — "
""
"splicing PP3 applies (supersedes missense BP4); "
"potential splice disruption."
""
).format(sa=spliceai)
elif splice_pp3_active:
splice_section = (
"
SpliceAI: {sa:.2f} — splicing PP3 applies"
).format(sa=spliceai)
else:
splice_section = "
SpliceAI: {sa:.2f}".format(sa=spliceai)
+ ps3_suppress_section = ""
+ if ps3_suppressed:
+ ps3_suppress_section = (
+ "
Note: {c} functional evidence not applied — "
+ "splicing PP3 takes precedence per CSpec."
+ ).format(c=ps3_suppressed)
+ af_divergent_section = ""
+ if af_divergent:
+ af_divergent_section = (
+ "
gnomAD AF (path-dependent): "
+ "the nt variants at this codon differ in "
+ "allele frequency; verify the exact change in the Allele Frequencies "
+ "track."
+ )
codon72 = ""
if codon == 72:
codon72 = (
"
Note: Codon 72 PS3/BS3 data is measured on the R72 "
"haplotype (rs1042522); reference is P72. P72X variants are NOT "
"in Table S3 (only R72X is). See description page for details."
)
ba1_section = ""
if ba1:
ba1_section = (
"
BA1 stand-alone Benign: "
"FAF ≥ 0.001 in gnomAD v4.1 forces class = Benign."
)
return (
"{warn}"
"
Variant: p.{name} (NP_000537.3)"
"
Provisional class: {cls} ({pts} pts)"
"
PM1: {pm1}"
"
PS3/BS3: {ps3}"
"
PP3/BP4: {pp3}"
"
AF (gnomAD v4.1): {af}"
"
BS2 (FLOSSIES): {bs2}"
"
Applied codes: {applied}"
- "{splice}{ba1_section}{codon72}"
+ "{splice}{ps3sup}{afdiv}{ba1_section}{codon72}"
"
NOT included in this sum: {cav}"
).format(warn=HEADER_WARNING,
name=name, cls=cls, pts=pts,
pm1=pm1 or "No contribution",
ps3=ps3 or "No contribution",
pp3=pp3 or "No contribution",
af=af_lbl or "No contribution",
bs2=bs2_lbl,
applied=applied or "(none)",
splice=splice_section,
+ ps3sup=ps3_suppress_section,
+ afdiv=af_divergent_section,
ba1_section=ba1_section,
codon72=codon72,
cav=CAVEATS_STR)
def generate_bed(s3, s2, hotspot_occ, af_lookup, present_set, flossies_lookup, tx):
lines = []
chrom = tx['chrom']
for (wt, codon, alt), s3_rec in sorted(s3.items(), key=lambda kv: (kv[0][1], kv[0][2])):
pm1_lbl, pm1_pts = pm1_code_and_points(wt, codon, alt, hotspot_occ)
ps3_lbl, ps3_pts = ps3_bs3_label_and_points(s3_rec['code'])
s2_rec = s2.get((wt, codon, alt))
pp3_lbl, pp3_pts = ('', 0)
spliceai = 0.0
+ spliceai_min = 0.0
paths = []
if s2_rec:
pp3_lbl, pp3_pts = pp3_bp4_label_and_points(s2_rec['code'])
spliceai = s2_rec.get('spliceai', 0.0)
+ spliceai_min = s2_rec.get('spliceai_min', spliceai)
paths = s2_rec.get('paths', [])
- # Megan's splicing rule: SpliceAI >= 0.2 -> PP3 splicing applies.
- # When the missense call is BP4 / BP4_Moderate, splicing PP3
- # supersedes — replace the BP4 contribution with PP3 (+1).
- splice_pp3_active = spliceai >= SPLICE_PP3_THRESHOLD
+ # Splicing rule (SpliceAI >= 0.2 -> PP3 splicing, superseding a missense
+ # BP4), applied only when UNAMBIGUOUS: every nt path yielding this protein
+ # change is >= threshold. SpliceAI is nt-specific; when the paths straddle
+ # the threshold this protein-level row cannot represent it, so we keep the
+ # missense code and flag the row for per-nt verification (Bioinformatic).
+ splice_all = bool(paths) and spliceai_min >= SPLICE_PP3_THRESHOLD
+ splice_straddle = (spliceai >= SPLICE_PP3_THRESHOLD
+ and spliceai_min < SPLICE_PP3_THRESHOLD)
+ splice_pp3_active = splice_all
splice_overrode_bp4 = False
if splice_pp3_active:
if pp3_lbl in ('BP4', 'BP4_Moderate'):
pp3_lbl = 'PP3 (splicing)'
pp3_pts = 1
splice_overrode_bp4 = True
elif not pp3_lbl:
pp3_lbl = 'PP3 (splicing)'
pp3_pts = 1
+ # CSpec caveat: do not apply a protein-level functional assay (PS3/BS3)
+ # when the variant is classified as splicing via PP3.
+ ps3_suppressed = ''
+ if splice_pp3_active and ps3_lbl:
+ ps3_suppressed = ps3_lbl
+ ps3_lbl, ps3_pts = ('', 0)
+
# Allele-frequency code (BA1 / BS1 / PM2_Supporting)
- af_code, af_pts, af_qty = af_code_and_points(
+ af_code, af_pts, af_qty, af_divergent = af_code_and_points(
wt, codon, alt, paths, af_lookup, present_set, tx)
af_lbl = "{} ({})".format(af_code, af_qty) if af_code else ''
ba1 = (af_code == 'BA1')
# BS2 from FLOSSIES (tiered by carrier count per CSpec GN009)
bs2_label, bs2_pts = bs2_evidence(wt, codon, paths, flossies_lookup, tx)
bs2_applies = bs2_label is not None
bs2_lbl = "{} ({} pts)".format(bs2_label, bs2_pts) if bs2_applies else "Not observed"
total = pm1_pts + ps3_pts + pp3_pts + af_pts + bs2_pts
if ba1:
cls = 'Benign'
else:
cls = bucket(total)
color = CLASS_COLOR[cls]
+ # nt-divergence flags: this protein-level row summarizes >1 nt variant
+ # whose SpliceAI or gnomAD AF disagree; verify the exact nt change in the
+ # Bioinformatic / Allele Frequency tracks.
+ nt_flags = []
+ if splice_straddle:
+ nt_flags.append('SpliceAI')
+ if af_divergent:
+ nt_flags.append('gnomAD AF')
+ nt_verify = " + ".join(nt_flags) if nt_flags else "-"
+
applied_codes = []
if pm1_lbl: applied_codes.append("{} (+{})".format(pm1_lbl, pm1_pts))
if ps3_lbl:
sign = "+" if ps3_pts > 0 else ""
applied_codes.append("{} ({}{})".format(ps3_lbl, sign, ps3_pts))
if pp3_lbl:
sign = "+" if pp3_pts > 0 else ""
applied_codes.append("{} ({}{})".format(pp3_lbl, sign, pp3_pts))
if af_code == 'BA1':
applied_codes.append("BA1 (stand-alone B)")
elif af_code:
sign = "+" if af_pts > 0 else ""
applied_codes.append("{} ({}{})".format(af_code, sign, af_pts))
if bs2_applies:
applied_codes.append("{} ({})".format(bs2_label, bs2_pts))
applied = "; ".join(applied_codes)
short = "{}{}{}".format(wt, codon, alt)
mo = mouseover(short, cls, total,
pm1_lbl, ps3_lbl, pp3_lbl, af_lbl, bs2_lbl, applied,
spliceai, splice_pp3_active, splice_overrode_bp4,
- codon, ba1)
+ codon, ba1, spliceai_min, splice_straddle, af_divergent,
+ ps3_suppressed)
segs = lib.aa_codon_genomic(codon, tx)
for g_start, g_end, _ex in segs:
lines.append("\t".join([
chrom, str(g_start), str(g_end),
short, "0", ".",
str(g_start), str(g_end),
color,
cls, str(total),
applied,
pm1_lbl or "-",
ps3_lbl or "-",
pp3_lbl or "-",
af_code or "-",
bs2_label if bs2_applies else "-",
"{:.2f}".format(spliceai) if spliceai else "0.00",
+ nt_verify,
mo,
]))
return lines
def build(db, outdir):
print("=== {} ===".format(db))
os.makedirs(outdir, exist_ok=True)
s3 = load_s3(SRC_S3)
s2 = load_s2(SRC_S2)
hotspots = load_hotspot_occurrences(HOTSPOTS_JSON)
af_lookup = load_af_lookup(db)
present_set = load_gnomad_present(db)
flossies_lookup = load_flossies_lookup(db)
print(" S3 entries: {} S2 entries: {} "
"cancerhotspots: {} AF: {} gnomAD present: {} FLOSSIES BS2: {}".format(
len(s3), len(s2), len(hotspots),
len(af_lookup), len(present_set), len(flossies_lookup)))
tx = lib.get_transcript_info(db)
bed_lines = generate_bed(s3, s2, hotspots, af_lookup, present_set, flossies_lookup, tx)
print(" {} BED rows".format(len(bed_lines)))
as_file = os.path.join(outdir, "TP53ProvisionalClass.as")
lib.write_autosql(as_file, AUTOSQL)
bed = os.path.join(outdir, "TP53ProvisionalClass_{}.bed".format(db))
with open(bed, 'w') as f:
f.write("\n".join(bed_lines) + "\n")
lib.run_sort_bed(bed)
bb = os.path.join(outdir, "TP53ProvisionalClass{}.bb".format(db.capitalize()))
- lib.run_bedToBigBed(bed, as_file, bb, lib.chrom_sizes_path(db), "bed9+10")
+ lib.run_bedToBigBed(bed, as_file, bb, lib.chrom_sizes_path(db), "bed9+11")
print(" wrote {}".format(bb))
from collections import Counter
cnt = Counter()
af_cnt = Counter()
bs2_cnt = Counter()
with open(bed) as f:
for line in f:
flds = line.split("\t")
cnt[flds[9]] += 1
af_cnt[flds[15]] += 1
if flds[16] in BS2_TIER_POINTS:
bs2_cnt[flds[16]] += 1
print(" Class distribution:")
for k, n in cnt.most_common():