GTEx読み会 #07

Effect of predicted protein-truncating genetic variants on the human transcriptome Rivas MA, et al.

Transcript

1. 2018/01/24 @օͰू·ͬͯGTEx࿦จΛҰؾʹྠಡ͢Δձ † Equally contributed (first, second last, and last) ‡ Equally contributed (2-4th authors) 2017 5: Quantifying…ͷlast author 2017 8: Landscape…ͷ4th authorͳͲ࠷௿3ใ 2015 3: Assesing…ͷ1st Broad Oxford U Helsinki U Washington Broad
2. • Genome sequencing (variantΛ֬ఆ) + RNA-Seq (࣮ࡍͷtranscriptʹ൓ө͞Εͯ ͍Δ͔ʁ) • λϯύΫ࣭Λ୹ॖͤ͞Δଟܕͷղੳ • ࡉ๔छಛҟੑ͸͋Δ͔ʁ • εϓϥΠγϯάͷଟܕͷӨڹͷଌఆ • HeterozygoteͳҨ఻ࢠͷෆ׆ੑԽ͸Ҩ ఻ࢠྔิঈͷϞσϧʹద߹͠ͳ͔ͬͨ • ຊจ4ϖʔδɺαϓϦ81ϖʔδ • citation 93 Protein truncating variants (PTVs) —Genetic variants predicted to shorten the coding sequence of genes GTEx + Genetic European variation in Health and Disease
3. • v3ͰಘΒΕ͍ͯΔσʔλʹରͯ͠ਪఆ • GTEx portalͰҨ఻ࢠͰݕࡧ • →ԼͷτϥοΫʹ௥Ճ͞Ε͍ͯΔ • શσʔλ͸ੜσʔλؚΊGTEx portal or dbGaPʹdeposit͞Ε͍ͯΔ • ʢGeuvadis͸ArrayExpressʹʣ
4. Subjects of variants defined as PTV stop codon stop codon exon skip Өڹ͸େ͖͍͸͕ͣͩͳͥ͋·Γେ͖ͳ໰୊ʹͳΒͳ͍ͷͩΖ͏ʁ → Non-sense mediated decay (NMD), allele specific expression (ASE)
5. • What to know? • Why they often do not have major phenotypic consequences? • in spite of PTV abundance in healthy individuals • In most cases their precise molecular effect has not been characterized and in other cases show gain-of-function effects avg. per individual Subjects of variants defined as PTV
6. .FUIPE • 4छͷଟܕʹ෼ྨ • ΞϨϧಛҟతͳൃݱͷݕग़ (allele specific expression, ASE) • Ref. Assessing allele-specific… • Compare transcripts between the PTV and the non-PTV alleles within the same individual • Validated by mmPCR-seq (microfluidics-based multiplex PCR and deep sequencing) Rui Zhang, et al. 2014. • Ҩ఻ࢠྔิঈ (Dosage compensation) • Focus only on biallelic whole-gene deletions with strong experimental support and manual curation • ΤΩιϯδϟϯΫγϣϯͷಛఆ • Assess variants influences on splicing disruption
7. .FUIPE • MAMBA — Allele specific expression pattern (Rivas, Pirinen et al Bioinformatics 2013.) • GTM* model to compute the statistics (Assessing allele-specific…ࢀর) • Splice disruption: MCMC fitting to model (shown below) • NMD: random forest individual (38 sequence and genomic features) π=proportion of the PTVs belonging to splice disruption group γk=normal or disruption in k N(0, 1) for general population Let yk be the standardized splice junction quantification value of the PTV carrier k Pirinen, et al. 2015 Normal, moderate, strong -> five states -> prior by Dirichlet
8. Gencode v12 جຊRPKM -> PEER (probabilistic estimation of expression residuals)Ͱิਖ਼ʁ
9. VBQTL (PEER)
10. Other applications • XHMM — CNV detection (large deletion) • BWA and GATK — SNV and indel detection
11. ͭΒ͍ᮢ஋ઃఆ in αϓϦ
12. Depth-based filtering Total depth count > 150, ref allele count > 5, and non-ref allele count > 5 ΞϥΠϯϝϯτιϑτ STAR vs Tophat ʢखಈ֬ೝϓϩηεʣ ʢͭΒ͍ʣ
13. Variant caller UG vs HCʢͭΒ͍ʣ
14. /POTFOTF
    4/7
    BOE
    JOEFM
    GSBNFTIJGU
    DBVTJOH
    "4& • 50-bpϧʔϧʹΑͬͯNMD༠ಋ͕༧ଌ͞ΕΔసࣸ෺ • ڧ͍ASE͕ΈΒΕͨ • ಉҰαϯϓϧ಺ͰPTVͱnon-PTV allelesͷൃݱྔΛൺֱ
15. There is no NMD when: 1) the resulting premature termination codon is in the last exon -or- 2) the resulting premature termination codon is in the last 50 nucleotides in the second to last exon http://sift.bii.a-star.edu.sg/www/indels_help.html?mybuild=hs38
16. Corresponding control Case sampled to include only a single heterozygote individual per variant
17. "4&
    QSFEJDUJPO
    BOE
    DMBTTJpDBUJPO • ASEΛ΋ͱʹNMDͷ༧ଌϞσϧΛͭ͘Δ • Random forest + 50-bp rule ʢˡ͍͖͞ΐ͏ʣ • ASE-based annotation (no NMD=escape and others=trigger) • ༧ଌ͞ΕͨNMDΛ༠ಋ͢Δكͳมҟ (n = 287) 17.9% hetero vs 8.1% specific Heterogeneous: heterogenous effects for each tissue type Tissue-specific: specific to a single tissue 30.5% 48.3% 3.3% 38 features including 50-bp rule
19. Common vs rare variants
20. • E: having heterogeneous ASE effects across the six tissues • the nonsense variant rs149244943 in gene PHKB (phosphorylase kinase, beta) • F: classified as having moderate ASE across all tissues • the nonsense variant rs119455955, a disease mutation for recessive late- infantile neuronal ceroid lipofuscinosis in gene TPP1 (tripeptidyl peptidase I) "4&
    UJTTVF
    TQFDJpDJUZ
    ʢ࣮ྫʣ
21. • Ҩ఻ࢠྔิঈ (dosage compensation)Λߟྀ͠ͳ͍Ϟσϧͷํ͕fitͨ͠ • 11ͷكͳܽଛͷHeterozygousͳΩϟϦΞͷൃݱྔʹ༗ҙͳൃݱݮগ • ಉ༷ͳൃݱݮগ͕53 nonsense PTVs with strong ASE signalsʹ • suggest that full dosage compensation is rare for human genes %PTBHF
    DPNQFOTBUJPO
    FTUJNBUJPO
22. Dosage compensation model • Dosage compensation: dominant model • Not: additive https://www.researchgate.net/post/ How_can_I_calculate_the_additive_ and_dominance_effect_of_a_SNP_u sing_R_for_quantitative_trait
23. Somatic mosaicism (mosaic deletion is present only in blood) →lung, adipose, and muscle could be mistaken
24. Splicing disruption —more distant sites are also known to affect splicing —Variation around splice junctions tends to be rare (minor allele frequency <= 0.01) Reported in clinvar Swedish exome sequencing Median genomic evolutionary rate Branch site SDM ratio Proportion of variants inducing SDM Other evidences
25. • No evidence for widespread dosage compensation • + most human genes are haplosufficient • -> homeostatic mechanisms (possibly as proposed in the theory of dominance) in heterozygous and inactivation in homozygous • Larger data sets will be required to increase our power to predict molecular consequences of variants from sequence data alone • personal transcriptomics will become an important complement to genome analysis. • MAMBA contains many methods are shared (e.g. NMD predictor, splicing disruption model) ! %JTDVTTJPO
26. • ஫໨͍ͯ͠ΔҨ఻ࢠ্ͷଟܕ͕࣮ࡍʹλϯύΫ࣭ʹͲΜͳӨڹΛٴ΅͠ ͍ͯΔ͔ • ΞϨϧಛҟతͳҨ఻ࢠൃݱྔ • εϓϥΠγϯάΛมԽͤ͞Δ͔ • variantͱRNA-seqͷσʔλ͔Βදݱܕʹͭͳ͛Δํ๏࿦Λཱ֬ • Ҩ఻ࢠิঈͷϞσϧ͸Somatic mosaicismʹ΋஫ҙ • ΦϧλφςΟϒεϓϥΠγϯάͷϞσϧʢin silicoؚΉʣ͸͜ͷ࿦จ ͷલޙͰ΋ଓʑͱग़͍ͯΔ • ࠷৽ͷྲྀΕΛΩϟονΞοϓͨ͠ํ͕Α͍͔΋ • Aspedia, LINSIGHT φχχπΧΤϧʁ