Factors influencing success of clinical genome sequencing across a broad spectrum of disorders

Abstract

To assess factors influencing the success of whole-genome sequencing for mainstream clinical diagnosis, we sequenced 217 individuals from 156 independent cases or families across a broad spectrum of disorders in whom previous screening had identified no pathogenic variants. We quantified the number of candidate variants identified using different strategies for variant calling, filtering, annotation and prioritization. We found that jointly calling variants across samples, filtering against both local and external databases, deploying multiple annotation tools and using familial transmission above biological plausibility contributed to accuracy. Overall, we identified disease-causing variants in 21% of cases, with the proportion increasing to 34% (23/68) for mendelian disorders and 57% (8/14) in family trios. We also discovered 32 potentially clinically actionable variants in 18 genes unrelated to the referral disorder, although only 4 were ultimately considered reportable. Our results demonstrate the value of genome sequencing for routine clinical diagnosis but also highlight many outstanding challenges.

Figure 1. Overview of projects and results

For each disorder, the number of independent cases (bars) studied is shown alongside information about the nature of the disorder: familial disorders (category 1, light green triangles), severe single-generation disorders suspected to be caused by de novo or recessive mutations (category 2, dark green), unrelated sporadic disorders (category 3, light blue) and extreme cases of common complex diseases (category 4, dark blue). The proportion of cases with each outcome class A-E is also shown (see Online Methods): pathogenic variant in novel gene for disorder (A, red circles), pathogenic variant in gene for related disorder (B, brown), pathogenic variant in known gene for disorder (C, pink), candidate pathogenic variant with validation studies underway (D, orange) and no single candidate variant, or negative results for validation of top candidate/s (blue). Size of points proportional to outcome fraction. Disorders are ranked by fraction of cases with confirmed pathogenic variants (class A to C).

Figure 2. The burden of variants of unknown significance

(a) Histograms of the number of previously unreported coding variants at conserved positions in different sets of candidate gene (Tiers 1, 1+2 and 1+2+3 for columns left to right) for early-onset epilepsy, under different inheritance models, across 216 WGS500 samples. (b) Histogram of the number of previously unreported coding variants at conserved positions in known X-linked mental retardation genes (XLMR), for the 99 male WGS500 samples. The candidate genes were chosen by high-throughput searches (Online Methods). Sample identifiers indicate individuals with the disorder in question. Sample names in green text indicate that the variant is not likely to be pathogenic (since it does not fit a plausible inheritance model or is less functionally compelling than another candidate); blue text indicates that the variant is thought to be causal (see Supplementary Table 6). OTH: Ohtahara syndrome; EOE: nonsyndromic early onset epilepsy; MR: mental retardation. See Supplementary Fig. 4 for the analysis of craniosynostosis.

Figure 3. Identification of de novo HUWE1 mutation associated with severe craniosynostosis

(a) Upper panel, the proband (CRS_4659; female, aged 6 months) presented with an abnormal skull shape. Lower panel, three-dimensional CT scan aged 5 months shows multisuture synostosis with multiple craniolacunae. (b) Family pedigree showing dideoxy sequence chromatograms with de novo G>A mutation of the X-linked HUWE1 gene in the proband (red arrow). Schematic X chromosomes are annotated from top to bottom with the HUWE1 alleles, haplotype of AA/CC polymorphisms located 1.15 kb away from mutation and used to deduce paternal origin, and androgen receptor (AR) trinucleotide repeat allele size (allele sizes in CRS_4654 and CRS_5215 are in brackets to emphasize that phase is unknown relative to other parts of the two X chromosomes). Note that the HUWE1 mutation abolishes a HpaII restriction site. (c) Analysis of X-inactivation in whole blood samples at AR locus. For each individual, AR alleles are indicated by arrows in the upper panel, while the lower panel shows proportions of methylated alleles and percentage representation of the more highly inactivated X chromosome. (d) Exclusive expression of cDNA from the HUWE1 mutant allele in both fibroblast (Fib) and Epstein Barr virus (EBV)-transformed lymphoblastoid cells from the proband. Arrows highlight absence of expression of the normal allele in either cell type. Product sizes (bp) from different alleles are shown on the right. WT: wild-type, Mut: mutant. (e) X chromosome ideogram showing eight de novo mutations identified. Where known, the parental allele on which the variant arose is indicated.

Figure 4. Candidate pathogenic noncoding variants

(a) Multi-species alignment of a region of the 5′ UTR of EPO in which a variant was identified at a conserved position (red text) in two families with erythrocytosis. (b) Erythrocytosis pedigrees studied, showing affected individuals (shaded grey), those sequenced (red borders), and genotypes of all individuals for whom we had DNA. We had no information about the father of PAR09 (dotted box).(c) Summary of read mapping in an individual with hypoparathyroidism showing evidence for an interstitial insertion-deletion event in which a ~ 50 kb region of chromosome 2p25.3 (top panel) has been duplicated and inserted into chromosome X, resulting in a 1.4 kb deletion 81.5 kb downstream of SOX3 (bottom panel). Yellow reads: mate maps to chrX; red reads: mate maps to chr2; grey reads: read and mate map to the same chromosome; white reads: read has mapping quality 0. (d) Pedigree showing segregation of the complex variant within the affected pedigree, with PCR validation below. M: mutation; N: normal. Primers 2SPF and XSPR flank the distal breakpoint of the deletion-insertion and are shown in Supplementary Figure 8. Primers XSPF and XSPR detect the normal allele. The mutation was not seen in 150 alleles from 100 unrelated normocalcemic individuals (50 males and 50 females, including N1 and N2, who are shown).