# assets/genotype_rules.tsv: the genotype rules of step 42 (docs/42-traits.md#genotype-rules). # A rule reads several sites together: the two haplotypes their genotypes make give one result. # Hand-curated from the primary papers in the pmids column; no text is taken from SNPedia or any # other database. GRCh38: ref is the GRCh38 base and alt the other allele the rule reads, both on # the plus strand as in dbSNP's GRCh38 record, also for a gene on the minus strand. # scripts/ci/check-rule-alleles.sh checks every site against dbSNP; og traits check --table # assets/traits.tsv --rules assets/genotype_rules.tsv checks the rest (tests/test_traits_report.py # runs both). # Four kinds of line, tab separated, each with its own columns: # rule gene trait needs_phase ancestry note pmids # site rsid chrom pos ref alt (one per site, in the order the haplotypes use) # haplotype name alleles (one base per site, comma separated) # combination diplotypes label reading # diplotypes: two haplotype names in the order of their haplotype lines, joined by /; when the # genotypes allow several diplotypes, all of them in that order, joined by |. # needs_phase yes: with two or more heterozygous sites the result needs to know which alleles sit # on the same copy (WhatsHap, from the reads); without that the rule is unphased, not read. # needs_phase no: the combination line is the one of every diplotype the genotypes allow. # ancestry: yes when the reading depends on ancestry (the note says how). rule APOE APOE APOE alleles (e2, e3, e4) no yes Odds ratios from a meta-analysis of people of European descent (Farrer 1997); the e4 effect was weaker in African American and Hispanic people and stronger in Japanese people. Step 25's Alzheimer's score (PGS000334) includes the same two sites; docs/42-traits.md says why step 06 does not list APOE. 8346443,9343467 site APOE rs429358 chr19 44908684 T C site APOE rs7412 chr19 44908822 C T haplotype APOE e1 C,T haplotype APOE e2 T,T haplotype APOE e3 T,C haplotype APOE e4 C,C combination APOE e3/e3 e3/e3 e3/e3, the most common pair: the reference group for Alzheimer's risk. combination APOE e2/e3 e2/e3 e2/e3: odds of Alzheimer's disease about 0.6 times those of e3/e3. combination APOE e2/e2 e2/e2 e2/e2: odds of Alzheimer's disease about 0.6 times those of e3/e3 (not significant in that study). combination APOE e3/e4 e3/e4 One e4: odds of late-onset Alzheimer's disease about 3.2 times those of e3/e3. A risk factor, not a diagnosis. combination APOE e1/e3|e2/e4 e2/e4 e2/e4: odds of Alzheimer's disease about 2.6 times those of e3/e3. Without phase this pair could also be the rare e1/e3. A risk factor, not a diagnosis. combination APOE e4/e4 e4/e4 Two e4: odds of late-onset Alzheimer's disease about 14.9 times those of e3/e3, the highest. Still a risk factor, not a diagnosis. combination APOE e1/e4 e1/e4 A pair with the rare e1 allele: not read. combination APOE e1/e2 e1/e2 A pair with the rare e1 allele: not read. combination APOE e1/e1 e1/e1 A pair with the rare e1 allele: not read. rule TAS2R38 TAS2R38 Bitter taste haplotype (TAS2R38, PTC and PROP) yes no The three TAS2R38 sites of the traits table read together: PAV (Pro49, Ala262, Val296) is the taster form and AVI (Ala49, Val262, Ile296) the non-taster form. Kim 2003 found five haplotypes worldwide; this rule reads the two common ones, and a copy with another (AAV, AAI, PVI or a rarer one) is not read. TAS2R38 lies on the minus strand, so the alleles here are the complement of the codon bases. 12595690,15723792 site TAS2R38 rs713598 chr7 141973545 C G site TAS2R38 rs1726866 chr7 141972905 G A site TAS2R38 rs10246939 chr7 141972804 T C haplotype TAS2R38 PAV G,G,C haplotype TAS2R38 AVI C,A,T combination TAS2R38 PAV/PAV PAV/PAV Two copies of PAV, the taster form: in Bufe 2005 this group rated PTC and PROP the most bitter. combination TAS2R38 PAV/AVI PAV/AVI One PAV and one AVI copy: a taster. In Bufe 2005 this group recognised PTC and PROP as bitter at concentrations close to those of PAV/PAV; how bitter they rated them lay between the two homozygous groups and varied widely. combination TAS2R38 AVI/AVI AVI/AVI Two copies of AVI, the non-taster form: in Bufe 2005 the AVI receptor did not respond to PTC, and this group recognised PTC as bitter only at far higher concentrations. PROP responses varied more. rule APOL1 APOL1 APOL1 kidney risk variants (G1 read, G2 not read) no yes G1 is two missense changes that sit on one copy, S342G (rs73885319) and I384M (rs60910145), in perfect linkage disequilibrium in Genovese 2010, so they are read together without phase. G2 (rs71785313) is a 6-base deletion of N388 and Y389; this rule table reads single-base sites only, so this rule does not read G2 and a G2 copy reads as G0 here; the HTML report reads G2 from the sample's VCF beside this result (og summary collect). G1 and G2 never share a copy. In African Americans, two risk copies (G1 or G2) went with focal segmental glomerulosclerosis (odds ratio 10.5) and hypertension-attributed end-stage kidney disease (odds ratio 7.3), and one copy did not raise the risk (a recessive pattern). Both variants are common on African chromosomes and were absent from the European chromosomes studied. 20647424 site APOL1 rs73885319 chr22 36265860 A G site APOL1 rs60910145 chr22 36265988 T G haplotype APOL1 G0 A,T haplotype APOL1 G1 G,G combination APOL1 G0/G0 G0/G0 No G1 copy. G2 is not read here, so this says nothing about it: two G2 copies would still be the high-risk genotype. combination APOL1 G0/G1 G0/G1 One G1 copy. G2 is not read here: if the other copy carries G2, this is G1/G2, a high-risk genotype; with no G2 it is one risk copy, which did not raise the risk in Genovese 2010. combination APOL1 G1/G1 G1/G1 Two G1 copies: the high-risk genotype of Genovese 2010, which went with focal segmental glomerulosclerosis and hypertension-attributed kidney failure in African Americans. A risk factor, not a diagnosis. rule SERPINA1 SERPINA1 Alpha-1 antitrypsin Pi type (M, S, Z) no yes The two common deficiency alleles of SERPINA1: S (Glu264Val, rs17580) and Z (Glu342Lys, rs28929474). SERPINA1 lies on the minus strand, so the alleles here are the complement of the codon bases. M stands for a copy with neither S nor Z; other deficiency and null alleles are not read. The readings follow the ATS/ERS 2003 standards; a serum alpha-1 antitrypsin level confirms a deficiency. Z is most common in people of northern European descent. 14522813 site SERPINA1 rs17580 chr14 94380925 T A site SERPINA1 rs28929474 chr14 94378610 C T haplotype SERPINA1 M T,C haplotype SERPINA1 S A,C haplotype SERPINA1 Z T,T combination SERPINA1 M/M PiMM No S and no Z copy: the usual type, with a normal alpha-1 antitrypsin level unless a rarer allele is present. combination SERPINA1 M/S PiMS One S copy: a carrier, with a level close to normal. combination SERPINA1 M/Z PiMZ One Z copy: a carrier, with an intermediate alpha-1 antitrypsin level. Smoking is the main added risk to the lungs. combination SERPINA1 S/S PiSS Two S copies: a mildly lowered level, not usually a deficiency that raises the risk of disease. combination SERPINA1 S/Z PiSZ One S and one Z copy: a moderate deficiency, with a raised risk of emphysema, above all in smokers. Worth a serum level and a talk with a doctor. combination SERPINA1 Z/Z PiZZ Two Z copies: severe alpha-1 antitrypsin deficiency, with a raised risk of early emphysema and of liver disease. Worth a serum level and a talk with a doctor. rule HFE HFE HFE haemochromatosis genotype (C282Y, H63D) no yes The two HFE missense changes of Feder 1996: C282Y (rs1800562), found homozygous in 83% of 178 patients with hereditary haemochromatosis, and H63D (rs1799945). The two sites are about 2 kb apart and read without phase: heterozygous at both reads as C282Y/H63D, one change on each copy. Many people with a risk genotype never build up iron; ferritin and transferrin saturation tell. The secondary-findings list reports C282Y homozygotes too (assets/acmg_sf.tsv); this line is the reading of every genotype. C282Y is most common in people of northern European descent. 8696333 site HFE rs1800562 chr6 26092913 G A site HFE rs1799945 chr6 26090951 C G haplotype HFE wt G,C haplotype HFE C282Y A,C haplotype HFE H63D G,G combination HFE wt/wt wt/wt Neither C282Y nor H63D. combination HFE wt/C282Y wt/C282Y One C282Y copy: a carrier. One copy alone rarely leads to iron overload. combination HFE wt/H63D wt/H63D One H63D copy: a common carrier type that rarely leads to iron overload. combination HFE C282Y/C282Y C282Y/C282Y Two C282Y copies: the genotype of most people with HFE haemochromatosis. Not everyone with it builds up iron; ferritin and transferrin saturation tell, and iron overload is treatable when found early. combination HFE C282Y/H63D C282Y/H63D One C282Y and one H63D copy: a compound heterozygote, with a lower risk of iron overload than two C282Y copies. Ferritin and transferrin saturation tell. combination HFE H63D/H63D H63D/H63D Two H63D copies: rarely leads to iron overload on its own. rule GC GC Vitamin D-binding protein type (Gc1F, Gc1S, Gc2) no yes rs7041 (Asp432Glu) and rs4588 (Thr436Lys) together give the three common vitamin D-binding protein types: Gc1F (Asp, Thr), Gc1S (Glu, Thr) and Gc2 (Asp, Lys). GC lies on the minus strand, so the alleles here are the complement of the codon bases. In Powe 2013 the two sites explained 79.4% of the variation in the protein level and 9.9% of that of total 25-hydroxyvitamin D. Gc1F was the most common type in black participants and Gc1S in white ones. The monoclonal assay that study used was questioned in later letters, so the protein levels are those of that assay. A protein type, not a vitamin D deficiency. 24256378 site GC rs7041 chr4 71752617 A C site GC rs4588 chr4 71752606 G T haplotype GC Gc1F A,G haplotype GC Gc1S C,G haplotype GC Gc2 A,T combination GC Gc1F/Gc1F Gc1F/Gc1F Two Gc1F copies: in Powe 2013 this group had the lowest vitamin D-binding protein level (mean 93 micrograms per millilitre). combination GC Gc1F/Gc1S Gc1F/Gc1S One Gc1F and one Gc1S copy: in Powe 2013 this group had an intermediate protein level (mean 285 micrograms per millilitre). combination GC Gc1F/Gc2 Gc1F/Gc2 One Gc1F and one Gc2 copy: Powe 2013 gave protein levels for the homozygous groups and Gc1F/Gc1S only. combination GC Gc1S/Gc1S Gc1S/Gc1S Two Gc1S copies: in Powe 2013 this group had the highest vitamin D-binding protein level (mean 468 micrograms per millilitre). combination GC Gc1S/Gc2 Gc1S/Gc2 One Gc1S and one Gc2 copy: Powe 2013 gave protein levels for the homozygous groups and Gc1F/Gc1S only. combination GC Gc2/Gc2 Gc2/Gc2 Two Gc2 copies: in Powe 2013 this group had an intermediate protein level (mean 190 micrograms per millilitre). rule CAD9p21 CDKN2B-AS1 Chromosome 9p21 heart disease risk haplotype no yes Two tags of the 9p21 risk haplotype near CDKN2A and CDKN2B: rs10757278 (G, Helgadottir 2007, myocardial infarction) and rs1333049 (C, the strongest coronary artery disease signal of the WTCCC 2007 scan). McPherson 2007 found the same interval, with no link to cholesterol, blood pressure or diabetes. The two are read together without phase; a copy with only one of the two risk alleles is not read. The studies were of people of European descent. 17478679,17554300,17478681 site CAD9p21 rs10757278 chr9 22124478 A G site CAD9p21 rs1333049 chr9 22125504 G C haplotype CAD9p21 nonrisk A,G haplotype CAD9p21 risk G,C combination CAD9p21 nonrisk/nonrisk no risk copy No copy of the 9p21 risk haplotype. combination CAD9p21 nonrisk/risk one risk copy One copy of the 9p21 risk haplotype: in Helgadottir 2007 each risk copy raised the odds of myocardial infarction about 1.28 times. A common variant with a modest effect, not a diagnosis. combination CAD9p21 risk/risk two risk copies Two copies of the 9p21 risk haplotype, as in about 21% of the people Helgadottir 2007 studied: their estimated risk of myocardial infarction was 1.64 times that of non-carriers (McPherson 2007: about 30 to 40% higher risk of coronary heart disease). A common variant with a modest effect, not a diagnosis. rule MAPT MAPT 17q21.31 inversion lineage (H1, H2) no yes The 17q21.31 region around MAPT comes in two lineages, H1 and H2, that differ by a 900-kb inversion and do not recombine (Stefansson 2005); H2 was found in about 20% of European chromosomes, rarely in Africans and almost never in East Asians. Two H1/H2 tags are read together: rs1052553 in MAPT (A is H1, Wider 2010) and rs8070723 (Hoglinger 2011's proxy for H1/H2, its minor allele G in H2). A copy with H1 at one tag and H2 at the other is not read. GRCh38 carries H1. 15654335,19912324,21685912 site MAPT rs1052553 chr17 45996523 A G site MAPT rs8070723 chr17 46003698 A G haplotype MAPT H1 A,A haplotype MAPT H2 G,G combination MAPT H1/H1 H1/H1 Two H1 copies, the common lineage. In Hoglinger 2011 H1 was the risk lineage for progressive supranuclear palsy (95% of patient chromosomes against 77.5% of controls), and in Wider 2010 for Parkinson's disease (odds ratio 1.43 per H1 tag allele). Both diseases are rare, and most people with H1/H1 never develop them. combination MAPT H1/H2 H1/H2 One H1 and one H2 copy. In Stefansson 2005 H2 carrier women in Iceland had more children than non-carriers. combination MAPT H2/H2 H2/H2 Two H2 copies, the rarer lineage.