Abstract
Germline whole exome sequencing from molecular tumor boards has the potential to be repurposed to support clinical pharmacogenomics. However, accurately calling pharmacogenomics-relevant genotypes from exome sequencing data remains challenging. Accordingly, this study assessed the analytical validity of the computational tool, Aldy, in calling pharmacogenomics-relevant genotypes from exome sequencing data for 13 major pharmacogenes. Germline DNA from whole blood was obtained for 164 subjects seen at an institutional molecular solid tumor board. All subjects had whole exome sequencing from Ashion Analytics and panel-based genotyping from an institutional pharmacogenomics laboratory. Aldy version 3.3 was operationalized on the LifeOmic Precision Health Cloud with copy number fixed to two copies per gene. Aldy results were compared with those from genotyping for 56 star allele–defining variants within CYP2B6, CYP2C8, CYP2C9, CYP2C19, CYP2D6, CYP3A4, CYP3A5, CYP4F2, DPYD, G6PD, NUDT15, SLCO1B1, and TPMT. Read depth was >100× for all variants except CYP3A4∗22. For 75 subjects in the validation cohort, all 3393 Aldy variant calls were concordant with genotyping. Aldy calls for 736 diplotypes containing alleles assessed by both platforms were also concordant. Aldy identified additional star alleles not covered by targeted genotyping for 139 diplotypes. Aldy accurately called variants and diplotypes for 13 major pharmacogenes, except for CYP2D6 variants involving copy number variations, thus allowing repurposing of whole exome sequencing to support clinical pharmacogenomics.
Clinical pharmacogenetic (PGx) testing is used to optimize drug therapy by identifying individuals likely to receive drug benefit or experience adverse drug events. Clinical testing performed within Clinical Laboratory Improvement Amendments (CLIA)–certified diagnostic laboratories can include targeted genotyping of selected variants, full-gene sequencing, or next-generation sequencing (NGS) approaches, which include whole genome sequencing (WGS) and whole exome sequencing (WES).1 Most clinical laboratories currently use targeted approaches for clinical PGx testing, with validated genotyping panels being the most common method.2 However, NGS-based approaches are being used more commonly for clinical diagnostic testing because these methods provide vast amounts of genetic information and are becoming increasingly economical because of advances in sequencing technology. Extraction of PGx-relevant genotypes from NGS remains technically challenging, which has led to the development of several computational tools, including Cyrius,3 Stargazer,4 Astrolabe,5 StellarPGx,6 ImPGX,7 and Aldy,8 that have been validated for PGx genotype extraction from WGS or targeted sequencing. To date, the authors are not aware of any computational tool that has been validated for clinical PGx genotype extraction from WES.
WES is routinely used for several clinical applications, including within molecular tumor boards to guide targeted cancer therapy. Within the clinical workflow at tumor boards, WES is frequently obtained for both somatic (tumor) and germline (blood) specimens to identify actionable variants for cancer treatment selection.9 Although the therapeutic success of targeted cancer therapy is well documented,10 the corresponding germline sequencing also has potential to support PGx-guided therapeutic optimization for select cancer drugs (ie, fluoropyrimidine and thiopurine chemotherapies)11,12 and a host of supportive medications, including antidepressant and analgesic therapies.13,14 The potential clinical benefit of repurposing germline WES data to support multidisciplinary PGx has been confirmed by multiple investigations.15, 16, 17, 18 Therefore, establishing a validated method to reliably extract PGx information from germline WES data would have clinical utility not only for tumor board patients but also for other clinical populations in which WES is routinely obtained (eg, rare disease clinics).
The aim of this study was to adapt and analytically validate Aldy, a computational PGx genotyping tool, to support genotyping from clinical WES. Previously, Aldy has been validated to extract PGx genotypes from WGS and targeted sequencing.8,19 This study assessed Aldy's performance in determining genotypes for 13 major pharmacogenes using commercially obtained germline clinical WES for adult solid tumor patients seen at our institution's molecular tumor board. Aldy's diplotype and haplotype calls for these pharmacogenes were analytically validated via comparison to targeted genotyping results and Sanger sequencing from our institution's CLIA-certified PGx laboratory. Herein, Aldy's performance and caveats for use in supporting PGx genotyping from clinical WES data are presented.
Materials and Methods
Study Enrollment and Design
Patients with solid tumors seen at the Indiana University Precision Genomics Clinic (Indianapolis, IN) provided consent to enroll in the Indiana University Total Cancer Care Protocol (part of the Oncology Research Information Exchange Network-Wide Total Cancer Care initiative, https://www.oriencancer.org, last accessed January 13, 2022). Subjects agreed to have whole blood collected for pharmacogenetics genotyping performed at the Indiana University Pharmacogenomics Laboratory and clinical whole exome sequencing performed at Ashion Analytics (Phoenix, AZ) to be used for clinical and research purposes. A total of 164 patients were enrolled in this study from August 2019 to March 2020. The study protocol and the parent Total Cancer Care Protocol were approved by the Indiana University Institutional Review Board. All subjects provided written informed consent.
Panel-Based Genotyping
Clinical pharmacogenetic testing was performed on whole blood samples at the CLIA-certified, College of American Pathologists–accredited Indiana University Pharmacogenomics Laboratory using a validated custom-designed OpenArray platform (ThermoFisher, Waltham, MA). Genes included on the OpenArray platform, along with the number of unique variants tested for each gene, are as follows: CYP2B6 (2), CYP2C8 (3), CYP2C9 (6), CYP2C19 (7), CYP2D6 (13 and copy number testing for exon 9), CYP3A4 (2), CYP3A5 (3), CYP4F2 (1), DPYD (3), G6PD (2), NUDT15 (2), SLCO1B1 (1), and TPMT (3). A complete list of the variants that were tested for is available in Supplemental Table S1.
Germline Whole Exome Sequencing
Whole blood samples were sent to the CLIA-certified, College of American Pathologists–accredited Ashion Analytics Laboratory (Phoenix, AZ) for paired normal/tumor whole exome sequencing, commercially available as Ashion Genomic Enabled Medicine Exome Testing (GEM ExTra). Whole exome sequencing methods for GEM ExTra have been previously published.20 Briefly, germline DNA was isolated from whole blood samples and prepared using a custom xGEN target capture kit (Integrated DNA Technologies Inc., Coralville, IA) for library preparation. Libraries were sequenced using Illumina (San Diego, CA) HiSeq 2500 or NovaSeq 6000. Sequencing data were analyzed via a proprietary pipeline using validated bioinformatics tools (GEM ExTra pipeline 4.0). Reference genome used for alignment was National Center for Biotechnology Information GRCh37 or hg19. Both binary alignment map (BAM) and variant calling files were transferred into the Precision Health Cloud (LifeOmic, Indianapolis, IN).
Analysis of Whole Exome Sequencing Data with Aldy
A detailed description of Aldy's genotype extraction method has been previously published.8,21, 22 Aldy version 3.3 was downloaded (https://github.com/0xTCG/aldy, last accessed January 13, 2022) and operationalized in the Precision Health Cloud using JupyterLab notebooks. WES read depth for all variants tested on the genotyping panel or detected by Aldy was assessed for all 164 subjects using SAMtools23 in a JupyterLab notebook environment on BAM files. BAM files were used as input for Aldy to perform genotype extraction of PGx-relevant variants and diplotypes for CYP2B6, CYP2C8, CYP2C9, CYP2C19, CYP2D6, CYP3A4, CYP3A5, CYP4F2, DPYD, G6PD, NUDT15, SLCO1B1, and TPMT. Within the Precision Health Cloud, Aldy diplotype calls and read depth coverage for all BAM files were outputted in .csv file format for downstream analysis and validation.
Analytical Validation of Aldy
Panel-based genotyping results were used as the reference standard for analytical validation of Aldy within the study population of 164 subjects. This population was divided into a development cohort of 89 subjects and a validation cohort of 75 subjects. The development cohort was used to test Aldy's performance in extracting PGx genotypes from clinical WES data during iterative methodological adaptations to improve Aldy's accuracy. After the final adaptations were implemented, the validation cohort was used to assess Aldy's analytical performance.
Orthogonal Validation of Additional Alleles Detected by Aldy
Variants with actionable recommendations in Clinical Pharmacogenetics Implementation Consortium (CPIC) guidelines that were not included on our institutional genotyping panel but were detected by Aldy were orthogonally validated using PCR-based genotyping or Sanger sequencing. For PCR-based genotyping, commercially available TaqMan assays and reagents were used to assess selected variants. DNA was amplified by real-time PCR using the Quantstudio 12K Flex version 1.2.2 (ThermoFisher) and subjected to TaqMan allele discrimination using commercially available individual reagents. The assay identification numbers of individual reagents used are listed in Supplemental Table S2. Genotypes were determined by Genotyper version 1.3 (ThermoFisher) and diplotypes inferred by Alleletyper version 1.0 (ThermoFisher). Sanger sequencing of selected variants was performed using BigDye Terminator version 3.1 (ThermoFisher), followed by sequencing on Applied Biosystems 3500xL (ThermoFisher). Primers for selected genetic variants were designed using National Center for Biotechnology Information Primer-Blast (https://www.ncbi.nlm.nih.gov/tools/primer-blast, last accessed January 13, 2022) and the IDT OligoAnalyzer (Integrated DNA Technologies Inc.). Primer synthesis was performed by Integrated DNA Technologies Inc. Primer sequences for each selected variant are provided in Table 1. Chromatograms were analyzed using Mutation Surveyor version 4.0.7 (SoftGenetics, State College, PA).
Table 1.
Sequences of Sanger Sequencing Primers Used to Genotype for rs116855232 and rs746071566 (to Determine NUDT15∗2 and NUDT15∗9, Respectively)
| Variant | Direction | Sequence |
|---|---|---|
| rs116855232 | Forward | 5′-TGTAAAACGACGGCCAGTGCCTTTGTAAACTGGGCTTC-3′ |
| Reverse | 5′-CAGGAAACAGCTATGACCCAAATCTTCTCGGCCACCTA-3′ | |
| rs746071566 | Forward | 5′-TGTAAAACGACGGCCAGTCATTCCCCAACCTGATAGCC-3′ |
| Reverse | 5′-CAGGAAACAGCTATGACCCAACCGAGCCTTTCCTCTTC-3′ |
Results
Study Population
Aldy was adapted for use with clinical WES and validated within a population of 164 adult solid tumor patients seen at our institutional molecular tumor board who had germline WES data and genotyping performed at our CLIA- and College of American Pathologists–accredited institutional PGx laboratory. These subjects were divided into development (n = 89) and validation (n = 75) cohorts to ensure that rare genotypes were adequately represented in both cohorts. The demographic and clinical characteristics of the study population were as follows: 59 ± 13 (mean ± SD) years old; 51% male sex; 85% White race; and 99% non-Hispanic ethnicity. The most common primary cancer diagnoses were colorectal (13%), breast (12%), pancreatic (12%), prostate (9%), and soft tissue sarcoma (8%). These data are summarized in Supplemental Table S3, with stratification by cohort.
Methodological Development of Aldy to Support Genotype Extraction from WES
The development of Aldy and validation of its accuracy in extracting PGx genotypes from targeted sequencing data and whole genome sequencing data have been previously published.8,19 However, preliminary testing revealed several inaccuracies when the conventional Aldy tool was tested on WES data; these consisted of inaccurate copy number calls, inaccurate calls for diplotypes with ambiguous phasing, and inaccurate diplotype calls containing variants with poor sequencing coverage. To adapt Aldy for use with WES, the procedures listed below were implemented, and the adapted method was tested within the 89 subjects in the development cohort. These methodological adaptations are detailed in Figure 1.
Figure 1.
Schematic diagram of Aldy procedures along with incorporated methodological adjustments to support genotype extraction from whole exome sequencing (WES). BAM, binary alignment map; IP, integer programming; PGx, pharmacogenetic; SAM, sequence alignment map; SV, structural variation.
i) Applied a custom WES sequencing profile (WXS) and ii) fixed the gene copy number parameter to two copies for all genes. Aldy utilizes sequencing profiles to define the coverage structure of the data input for sequencing technologies that have uniform coverage patterns to detect copy number change and gene fusions. However, WES samples do not exhibit uniform coverage patterns and are unreliable for copy number calling.24 Therefore, Aldy's default whole genome sequencing profile (Illumina) was replaced with a new WXS profile that assumes a copy number of two for each gene to support genotype extraction from Ashion WES data in these analyses. Although this profile does not support detecting copy number changes or gene fusions from WES data, it nevertheless allows Aldy to accurately call genotypes for genes without these structural variations. Because of its location on the X chromosome, males have haploid genotypes for G6PD. To account for this, all Aldy G6PD calls in male subjects were confirmed to be homozygous (for wild-type or variant alleles) and converted to the corresponding hemizygous genotypes.
iii) Excluded subjects with CYP2D6 copy number variation (CNV). Because the copy number parameter was fixed to two, any subjects who had CYP2D6 CNV in their PGx laboratory results were excluded from Aldy validation, including 12 subjects from the development cohort and 11 subjects from the validation cohort. Because CYP2D6 deletions (ie, the ∗5 allele) and duplications are relatively common and affect clinical actionability, the inability to resolve CNV limits Aldy's utility to genotype CYP2D6 as a stand-alone tool. As an exploratory analysis, results from a CYP2D6 copy number assay performed in our institutional PGx laboratory (using real-time PCR targeting exon 9) were incorporated along with Aldy calls for the 23 subjects with CNV. When combining Aldy output and copy number results, CYP2D6 genotypes for all subjects were able to be resolved.
iv) Added warning for multiple optimal major star allele solutions. A warning was added to the Aldy output to identify cases in which multiple major star allele solutions were found. These cases were manually reviewed, and it was determined that the patients' diplotypes for these genes were unclear because of ambiguous phasing, which is a common limitation of short-read sequencing technologies. Therefore, these subjects were excluded from the diplotype concordance analysis only (they were still included in the variant-level concordance analysis) on a per-gene basis. The genes affected by ambiguous phasing in the validation cohort included CYP2D6 (ambiguous diplotypes for two subjects) and TPMT (ambiguous diplotypes for three subjects).
v) Excluded variants with read depth <30×. Read depth was assessed from BAM files for the 56 star allele–defining variants included in this analysis, with results shown in Supplemental Table S4. With the exception of CYP3A4∗22, which resides nearly 200 bp into an intron, read depth was ≥100× for all analyzed variants in all subjects' BAM files. Read depth at CYP3A4∗22 was <30× for 22 and 16 subjects in the development and validation cohorts, respectively, and these subjects were excluded from the CYP3A4 analyses.
After implementing these adaptations, the accuracy of the modified Aldy tool in extracting PGx genotypes from WES data was retested in 89 subjects within the development cohort. Aldy calls were concordant with orthogonal results from our PGx laboratory for all 3790 variant-level results and 911 diplotype-level results (diplotype-level results shown in Supplemental Table S5).
Analytical Validation of Modified Aldy Method
After implementing the adaptations described above, the analytical validity of Aldy was assessed within the validation cohort (n = 75 subjects). Aldy was assessed at the variant and diplotype levels across 56 star allele–defining variants in 13 pharmacogenes via comparison to results from panel-based genotyping performed in our institutional PGx laboratory. Aldy's variant-level results were completely concordant with those obtained from genotyping, with point estimates for accuracy, analytical sensitivity, and analytical specificity of 100% for all assessed variants (Table 2). In addition, Aldy's interassay concordance (ie, the ability to reproduce the same results for the same sequencing file over subsequent runs) and intra-assay concordance (ie, the ability to reproduce the same results for the same sequencing file multiple times within the same run) were both confirmed to be 100% for all variants. Table 3 displays the diplotype-level concordance of Aldy's results relative to results from genotyping. Aldy's calls were concordant to those from genotyping for all 736 diplotypes that included alleles assessed by both methods. In addition, Aldy detected additional major star alleles in 139 diplotypes, which included nine unique star alleles with actionable CPIC recommendations, which were not assessed in our genotyping assay. Subject-level results at the variant and diplotype levels are displayed in Supplemental Tables S6 and S7, respectively.
Table 2.
Variant-Level Orthogonal Validation of Aldy
| Variant | Variant allele count in development cohort (total chromosomes) |
Variant allele count in validation cohort (total chromosomes) |
Analytical parameters assessed in validation cohort |
||||
|---|---|---|---|---|---|---|---|
| Accuracy, % |
Analytical sensitivity, % (95% CI) |
Analytical specificity, % (95% CI) |
Interassay concordance, % |
Intra-assay concordance, % |
|||
| CYP2B6∗6 (c.516G>T) | 48 (178) | 42 (150) | 100 | 100 (92–100) | 100 (97–100) | 100 | 100 |
| CYP2B6∗8 (c.415A>G) | 1 (178) | 1 (150) | 100 | 100 (21–100) | N/A† | 100 | 100 |
| CYP2B6∗18 (c.983T>C) | 1 (178) | 1 (150) | 100 | 100 (21–100) | 100 (97–100) | 100 | 100 |
| CYP2C8∗2 (c.805A>T) | 0 (76) | 0 (40) | 100 | N/A | 100 (91–100) | 100 | 100 |
| CYP2C8∗3 (c.416G>A) | 9 (76) | 3 (40) | 100 | 100 (44–100) | 100 (91–100) | 100 | 100 |
| CYP2C8∗4 (c.792C>G) | 3 (76) | 1 (40) | 100 | 100 (21–100) | 100 (91–100) | 100 | 100 |
| CYP2C9∗2 (c.430C>T) | 30 (178) | 20 (150) | 100 | 100 (84–100) | 100 (97–100) | 100 | 100 |
| CYP2C9∗3 (c.1075A>C) | 10 (178) | 9 (150) | 100 | 100 (70–100) | 100 (97–100) | 100 | 100 |
| CYP2C9∗5 (c.1080C>G) | 1 (178) | 1 (150) | 100 | 100 (21–100) | 100 (97–100) | 100 | 100 |
| CYP2C9∗6 (c.818del) | 0 (178) | 0 (150) | 100 | N/A | 100 (98–100) | 100 | 100 |
| CYP2C9∗8 (c.449G>A) | 2 (178) | 0 (150) | 100 | N/A | 100 (98–100) | 100 | 100 |
| CYP2C9∗11 (c.1003C>T) | 0 (178) | 0 (150) | 100 | N/A | 100 (98–100) | 100 | 100 |
| CYP2C19∗2 (c.681G>A) | 26 (178) | 15 (150) | 100 | 100 (80–100) | 100 (97–100) | 100 | 100 |
| CYP2C19∗3 (c.636G>A) | 0 (178) | 0 (150) | 100 | N/A | 100 (98–100) | 100 | 100 |
| CYP2C19∗4 (c.1A>G) | 0 (178) | 0 (150) | 100 | N/A | 100 (98–100) | 100 | 100 |
| CYP2C19∗6 (c.395G>A) | 0 (178) | 0 (150) | 100 | N/A | 100 (98–100) | 100 | 100 |
| CYP2C19∗8 (c.358T>C) | 2 (178) | 0 (150) | 100 | N/A | 100 (98–100) | 100 | 100 |
| CYP2C19∗10 (c.680C>T) | 0 (178) | 0 (150) | 100 | N/A | 100 (98–100) | 100 | 100 |
| CYP2C19∗17 (c.-806C>T) | 33 (178) | 35 (150) | 100 | 100 (90–100) | 100 (97–100) | 100 | 100 |
| CYP2C19∗24 (c.1004G>A) | 1 (178) | 0 (150) | 100 | N/A | N/A† | 100 | 100 |
| CYP2D6∗2 (c.886C>T) | 61 (144) | 46 (128) | 100 | 100 (92–100) | 100 (96–100) | 100 | 100 |
| CYP2D6∗2 (c.1457G>C) | 86 (144) | 76 (128) | 100 | 100 (95–100) | 100 (93–100) | 100 | 100 |
| CYP2D6∗3 (c.775del) | 2 (144) | 2 (128) | 100 | 100 (34–100) | 100 (97–100) | 100 | 100 |
| CYP2D6∗4 (c.506-1G>A) | 23 (144) | 28 (128) | 100 | 100 (88–100) | 100 (96–100) | 100 | 100 |
| CYP2D6∗6 (c.454del) | 3 (144) | 1 (128) | 100 | 100 (21–100) | 100 (97–100) | 100 | 100 |
| CYP2D6∗7 (c.971A>C) | 1 (144) | 0 (128) | 100 | N/A | 100 (97–100) | 100 | 100 |
| CYP2D6∗8 (c.505G>T) | 0 (144) | 0 (128) | 100 | N/A | 100 (97–100) | 100 | 100 |
| CYP2D6∗9 (c.841_843del) | 4 (144) | 5 (128) | 100 | 100 (57–100) | 100 (97–100) | 100 | 100 |
| CYP2D6∗10 (c.100C>T) | 25 (144) | 30 (128) | 100 | 100 (89–100) | 100 (96–100) | 100 | 100 |
| CYP2D6∗14 (c.505G>A) | 0 (144) | 0 (128) | 100 | N/A | 100 (97–100) | 100 | 100 |
| CYP2D6∗15 (c.137dup) | 1 (144) | 1 (128) | 100 | 100 (21–100) | N/A† | 100 | 100 |
| CYP2D6∗17 (c.320C>T) | 3 (144) | 3 (128) | 100 | 100 (44–100) | 100 (97–100) | 100 | 100 |
| CYP2D6∗29 (c.1012G>A) | 2 (144) | 0 (128) | 100 | N/A | 100 (97–100) | 100 | 100 |
| CYP2D6∗41 (c.985+39G>A) | 18 (144) | 14 (128) | 100 | 100 (78–100) | 100 (97–100) | 100 | 100 |
| CYP2D6∗59 (c.975G>A) | 1 (144) | 2 (128) | 100 | 100 (34–100) | N/A† | 100 | 100 |
| CYP2D6∗62 (c.1321C>T) | 0 (144) | 1 (128) | 100 | 100 (21–100) | N/A† | 100 | 100 |
| CYP3A4∗2 (c.664T>C) | 0 (178) | 0 (150) | 100 | N/A | 100 (98–100) | 100 | 100 |
| CYP3A4∗22 (c.522-191C>T) | 8 (134) | 5 (118) | 100 | 100 (57–100) | 100 (97–100) | 100 | 100 |
| CYP3A5∗3 (c.219-237A>G) | 154 (178) | 129 (150) | 100 | 100 (97–100) | 100 (85–100) | 100 | 100 |
| CYP3A5∗6 (c.624G>A) | 2 (178) | 3 (150) | 100 | 100 (44–100) | 100 (97–100) | 100 | 100 |
| CYP3A5∗7 (c.1035dup) | 1 (178) | 0 (150) | 100 | N/A | 100 (98–100) | 100 | 100 |
| CYP4F2∗3 (c.1297G>A) | 41 (178) | 40 (150) | 100 | 100 (91–100) | 100 (97–100) | 100 | 100 |
| DPYD∗2 (c.1905+1G>A) | 0 (178) | 0 (150) | 100 | N/A | 100 (98–100) | 100 | 100 |
| DPYD∗13 (c.1679T>G) | 0 (90) | 0 (136) | 100 | N/A | 100 (97–100) | 100 | 100 |
| DPYD HapB3 (c.1236G>A) | 2 (90) | 0 (136) | 100 | N/A | N/A† | 100 | 100 |
| DPYD D949V (c.2846A>T) | 3 (90) | 1 (136) | 100 | 100 (21–100) | 100 (97–100) | 100 | 100 |
| G6PD A- (c.202G>A) | 3 (134) | 0 (111) | 100 | N/A | 100 (98–100) | 100 | 100 |
| G6PD A- (c.968T>C) | 1 (134) | 0 (111) | 100 | N/A | N/A† | 100 | 100 |
| G6PD A (c.376A>G) | 8 (134) | 4 (111) | 100 | 100 (51–100) | 100 (97–100) | 100 | 100 |
| NUDT15∗3 (c.415C>T) | 2 (30) | 1 (134) | 100 | 100 (21–100) | 100 (97–100) | 100 | 100 |
| NUDT15∗5 (c.52G>A) | 0 (30) | 0 (134) | 100 | N/A | 100 (97–100) | 100 | 100 |
| NUDT15∗9 (c.38GAGTCG[2]) | 2 (30) | 0 (134) | 100 | N/A | N/A† | 100 | 100 |
| SLCO1B1∗5 (c.521T>C) | 28 (178) | 16 (150) | 100 | 100 (81–100) | 100 (97–100) | 100 | 100 |
| TPMT∗2 (c.238G>C) | 0 (178) | 0 (150) | 100 | N/A | 100 (98–100) | 100 | 100 |
| TPMT∗3A,∗3B (c.460G>A) | 9 (178) | 5 (150) | 100 | 100 (57–100) | 100 (97–100) | 100 | 100 |
| TPMT∗3A,∗3C (c.719A>G) | 9 (178) | 5 (150) | 100 | 100 (57–100) | 100 (97–100) | 100 | 100 |
N/A, not applicable.
Analytical specificity was not calculated for these alleles because orthogonal validation was only performed on subjects having the allele based on Aldy results.
Table 3.
Diplotype-Level Concordance of Aldy with Orthogonal Methods in Validation Cohort
| Gene | Concordant diplotypes |
Detection of additional major star alleles |
Discordant diplotypes |
Diplotypes with CNV |
Indeterminate diplotypes |
Total diplotypes compared |
|---|---|---|---|---|---|---|
| CYP2B6 | 47 | 28 | 0 | 0 | 0 | 75 |
| CYP2C8 | 20 | 0 | 0 | 0 | 0 | 20 |
| CYP2C9 | 74 | 1 | 0 | 0 | 0 | 75 |
| CYP2C19 | 74 | 1 | 0 | 0 | 0 | 75 |
| CYP2D6 | 50 | 12 | 0 | (11) | (2) | 62 |
| CYP3A4† | 57 | 2 | 0 | 0 | 0 | 59 |
| CYP3A5 | 75 | 0 | 0 | 0 | 0 | 75 |
| CYP4F2 | 54 | 21 | 0 | 0 | 0 | 75 |
| DPYD | 22 | 53 | 0 | 0 | 0 | 75 |
| G6PD | 75 | 0 | 0 | 0 | 0 | 75 |
| NUDT15 | 61 | 1 | 0 | 0 | 0 | 62 |
| SLCO1B1 | 56 | 19 | 0 | 0 | 0 | 75 |
| TPMT | 71 | 1 | 0 | 0 | (3) | 72 |
| Total | 736 | 139 | 0 | (11) | (5) | 875 |
Detection of additional alleles with the same tag variant (eg, CYP2B6∗6 and CYP2B6∗9, SLCO1B1∗5 and SLCO1B1∗15) was not considered as discordant or as detection of additional major star alleles within these analyses. Values in parentheses were not included in calculations because subjects with copy number variation or indeterminate diplotypes were excluded from concordance analyses for the relevant genes.
CNV, copy number variation.
Sixteen subjects were excluded from the diplotype analysis based on having CYP3A4∗22 read depth <30×.
Identification and Validation of Additional CPIC-Actionable Alleles Detected by Aldy that Were Not Covered by Panel-Based Genotyping
The additional alleles detected by Aldy that were not covered by our genotyping assay included nine alleles with actionable CPIC recommendations, consisting of low minor allele frequency variants in CYP2B6, CYP2C19, CYP2D6, DPYD, G6PD, and NUDT15, as summarized in Table 4. Aldy calls were confirmed using orthogonal methods, including PCR-based genotyping and Sanger sequencing, for all 14 subjects possessing these alleles, with results shown in Table 2. Analytical specificity was not assessed for these alleles because orthogonal methods were only performed on patients with the relevant variants, as detected by Aldy.
Table 4.
Additional CPIC-Actionable Variants Detected by Aldy that Were Not Assessed by Panel-Based Genotyping Reference Standard
| Gene | Variant information: rs ID (associated alleles) |
Functional effect† | Variant allele count in development and validation cohorts (total chromosomes) |
Minor allele frequency‡ |
||
|---|---|---|---|---|---|---|
| African | American admixed | European | ||||
| CYP2B6 | rs12721655 (∗8, ∗13) | No function | 2 (328) | <0.001 | — | 0.004 |
| CYP2C19 | rs118203757 (∗24) | No function | 1 (328) | <0.001 | — | <0.001 |
| CYP2D6 | rs774671100 (∗15) | No function | 2 (282) | <0.001 | 0.002 | <0.001 |
| rs79292917 (∗59) | Decreased | 3 (282) | 0.002 | — | 0.005 | |
| rs730882251 (∗62) | No function | 1 (282) | <0.001 | — | <0.001 | |
| DPYD | rs56038477 (HapB3)§ | Decreased | 2 (226) | 0.003 | — | 0.020 |
| G6PD | rs76723693 (A-) | Deficient | 1 (245) | 0.001 | — | <0.001 |
| NUDT15 | rs746071566 (∗2, ∗9) | No function | 2 (164) | <0.001 | — | 0.001 |
—, Data not available; CPIC, Clinical Pharmacogenetics Implementation Consortium; ID, identifier.
Functional effects for each variant are defined by the Pharmacogene Variation Consortium (http://pharmvar.org, last accessed January 13, 2022) or the CPIC (http://cpicpgx.org, last accessed January 13, 2022).
Minor allele frequencies are based on variant frequency data from the Allele Frequency Aggregator project, sponsored by the National Center for Biotechnology Information for all variants except rs12769205. For rs12769205, we report the frequency of ∗35 allele rather than the rs12769205 variant because rs12769205 is also contained in the more common CYP2C19∗2 allele.
The coding variant rs56038477 was used to tag rs75017182 and the HapB3 haplotype.
Discussion
This investigation details the analytical validation of a computational method to accurately extract clinically actionable PGx information for 13 pharmacogenes from WES data obtained during the clinical workflow of our institutional molecular tumor board. After implementing a series of methodological changes to adapt Aldy for use with WES data, complete concordance was found between Aldy's genotype calls and those from validated orthogonal methods at both the individual variant and diplotype levels. Aldy also detected nine additional low-frequency, clinically actionable alleles that were not included in our institutional genotyping platform but were confirmed via orthogonal testing. The variants in the validation analyses included single-nucleotide variations and small insertions, deletions, and duplications, confirming Aldy's ability to accurately detect multiple types of genetic variation. These findings demonstrate that Aldy is an accurate and efficient tool to repurpose clinical WES data to support multidisciplinary PGx.
Although the clinical potential of repurposing clinical NGS data to support PGx has long been recognized, significant technical challenges have limited the feasibility of accurate genotype extraction. Recently, bioinformatic approaches have been developed to overcome these challenges to enable accurate genotyping of major pharmacogenes from NGS. These approaches include computational tools to extract PGx genotypes from NGS sequencing files, including BAM files (used as input for Aldy, Astrolabe, Cyrius, Stargazer, ImPGx, and StellarPGx) and variant calling files (used as input for PharmCAT). Evaluations of the analytical performance of these tools have demonstrated accurate extraction of PGx genotypes from WGS and targeting sequencing data, although structural variation in CYP2D6 remains challenging.19 However, given inherent challenges related to WES data, including nonuniform genomic coverage and omission of clinically actionable variants in noncoding genomic regions, these methods had not been previously adapted to perform accurate PGx genotyping from WES. An investigation by van der Lee et al18 successfully extracted PGx genotypes from WES for 39 PGx-relevant variants using the GenomeAnalysisToolkit HaplotypeCaller, although significant limitations of their method included the inability to reliably extract genotypes for important pharmacogenes, like CYP2C19 and CYP3A5. Therefore, these findings are novel in that they validate a genotype extraction tool for use with WES data and expand the existing capability of genotype extraction from WES to include additional variants and actionable pharmacogenes.
By enabling accurate genotyping from WES data, Aldy can be effectively used to repurpose genetic information obtained from existing clinical workflows to support PGx for actionable pharmacogenes, including detection of low-frequency variants that are not typically assessed using panel-based approaches.25 Within our health system, WES has become the preferred NGS format for many clinical services that require large-scale genetic data because it provides a large amount of clinically actionable information at a lower cost and with reduced computational resource requirements relative to WGS. Therefore, we plan to integrate genotypes extracted by Aldy for validated pharmacogenes into patient electronic health records, obviating the need for panel-based genotyping to obtain PGx information. It is anticipated that Aldy-based PGx implementation will have significant clinical utility because past investigations have found that approximately 60% of patients seen at molecular tumor boards are prescribed at least one medication included in a CPIC guideline.17,26 Similarly, past research has identified abundant opportunities for PGx in patients who had WES performed to aid in diagnosis of suspected genetic diseases, which constitutes another practical population for Aldy-based PGx implementation.15,18 As technological advances continue to reduce the cost and turnaround time for clinical WES sequencing, more patients should have WES data available that will, in turn, allow increased opportunities for PGx implementation using Aldy.
Although this study validates an Aldy-based tool to accurately extract genotypes from WES, the following caveats must be appreciated when implementing Aldy for use with WES data: i) Aldy cannot accurately determine gene copy number from WES data. Although fixing the copy number at two gene copies enables accurate genotyping of pharmacogenes without CNV, it does not allow for genotyping of genes in which CNV is common, which is an established limitation of current genotype extraction methods using WES data.24 Therefore, Aldy output is insufficient to accurately determine CYP2D6 genotypes using WES data because it is unable to satisfy the minimum requirements for CYP2D6 testing recommended by the Association for Molecular Pathology.27 This study's findings suggest that Aldy results may be used in combination with validated methods for determining CYP2D6 copy number28,29 to accurately determine CYP2D6 genotypes. However, strategies integrating Aldy results with those from established orthogonal methods to detect CYP2D6 copy number variations (eg, quantitative PCR and multiplex-dependent probe amplification) need to be analytically validated before use. ii) Aldy does not currently support phasing of detected variants. Similarly, the WES data from Ashion Analytics used in this validation did not include a way to conclusively determine phasing. Therefore, Aldy could not resolve three ambiguous diplotypes that were encountered in CYP2B6 (∗1/∗13 versus ∗6/∗8), CYP2D6 (∗1/∗4 versus ∗4/∗10), and TPMT (∗1/∗3A versus ∗3B/∗3C), which were also not distinguishable by our panel-based genotyping reference standard. However, the implementation of a warning within the Aldy output may be used to label diplotype solutions in which ambiguous phasing was encountered so that alternative methods can be used to accurately resolve the genotype or phenotype. iii) Aldy can only accurately genotype variants that are adequately covered by WES, which is generally recommended to be ≥30× for NGS data.30 Within the development cohort, one discordant result was found between Aldy and the genotyping reference standard for a subject with 20× coverage at CYP3A4∗22. When excluding subjects with <30× coverage at any locus in the validation analyses, all Aldy calls were concordant with reference standards. The authors were unable to obtain Aldy results or assess the analytical performance of Aldy for genotyping PGx-relevant noncoding variants that were not covered by Ashion Analytics GEM ExTra (eg, VKORC1 –1639G>A). WES coverage is determined by the target capture probes utilized for DNA library preparation, which can vary based on the clinical sequencing vendor. For example, GEM ExTra sequencing provided sufficient coverage of numerous noncoding PGx variants, including CYP2C19∗17, CYP3A5∗3, and CYP3A4∗22 (in most patients). Determining that there is sufficient coverage of the PGx variants of interest within the WES or other Aldy data input is a necessary step to ensure the accuracy of Aldy's results. iv) Aldy only detects variants included within its gene database files and does not identify novel variants. Aldy gene database files are regularly updated based on curations from expert professional PGx organizations, including the Pharmacogene Variation Consortium and the Pharmacogenomics Knowledge Base, and therefore contain complete lists of clinically relevant PGx variants, including those with low allele frequencies that are not practical to assess via targeted genotyping approaches. For SLCO1B1, the current star allele definitions have not been fully characterized, resulting in Aldy calling extra variants in addition to defined haplotypes for many subjects. However, Pharmacogene Variation Consortium SLCO1B1 gene experts are currently updating star allele definitions for standardization.31 Once released, Aldy's SLCO1B1 gene database will be revised with the updated star allele definitions from Pharmacogene Variation Consortium. Aldy gene databases can also be customized by users to include additional variants of interest. As of version 3.3, Aldy supports genotype extraction for 15 of 19 pharmacogenes with CPIC recommendations, with CACNA1S, HLA-A, HLA-B, and RYR1 not being supported.
This study has several limitations. First, the primary samples for comparison, which consisted of panel-based genotyping results from our institutional CLIA- and College of American Pathologists–accredited PGx laboratory, were only available for 46 relatively common functional variants within the 13 pharmacogenes assessed in the validation (variant list shown in Supplemental Table S1). To account for this, orthogonal validation via PCR-based genotyping or Sanger sequencing to assess the accuracy of Aldy calls for less common CPIC-actionable variants not included on the genotyping panel was performed. Although these methods allowed us to assess the sensitivity of Aldy genotyping for these variants, Aldy's specificity was not assessed because, due to practical considerations, orthogonal testing was only performed in subjects identified by Aldy as having these variants. Next, although the cohort size was sufficient to validate Aldy for most CPIC-actionable variants, rare CPIC-actionable variants were not able to be validated because they were not detected in the cohort. Also, the fact that most of the study population was White limited the ability to validate alleles that are specific to other racial groups (eg, CYP3A5∗7). In addition, the study population was smaller for variants in more recently characterized pharmacogenes (eg, NUDT15) because these variants were more recently added to our genotyping panel and, as a result, there were fewer subjects for orthogonal comparison. Finally, CPIC guidelines currently provide recommendations for atazanavir therapy based on UGT1A1 genotype, and Aldy version 3.3 supports genotyping extraction for UGT1A1. However, because atazanavir is not widely prescribed at our institution, UGT1A1 is not included on our institution's PGx genotyping panel. Because of a lack of samples for comparison, Aldy genotype extraction was not assessed for UGT1A1.
In conclusion, this study demonstrates the analytical validity and caveats for use of a computational method to accurately extract genotypes for 13 major pharmacogenes from clinical WES data obtained for patients seen at our institutional molecular tumor board. Given the abundant opportunities for PGx in patients with advanced cancer, it is expected that employing Aldy to repurpose WES data to support multidisciplinary PGx will have broad clinical benefits. Future directions of this research will involve the following: analytical validation of Aldy for use with clinical WGS data and WES data from other laboratories and informatic pipelines; integrating copy number assay and Aldy results to enable clinically valid CYP2D6 genotyping from WES; and incorporating validated Aldy genotype results into electronic health records to promote genotype-guided prescribing.
Acknowledgments
We thank Elizabeth B. Medeiros, M.S., and Caitlin A. Granfield, M.S., for performing orthogonal analyses to validate variants identified by Aldy that were not included on our institutional genotyping panel.
Footnotes
Supported with funding from NIH grant R35GM131812 (T.C.S.) and the Indiana University Grand Challenge Precision Health Initiative (R.C.L., T.S., M.R., B.P.S., and T.C.S.).
R.C.L. and T.S. contributed equally to this work.
Disclosures: Some authors are current or former employees of LifeOmic Inc. (R.R., S.M.B., B.A.S., and B.P.), a commercial entity that supports management and analysis of genetic data; and Caris Life Sciences (M.R.), a commercial entity that performs clinical sequencing services. V.M.P. is laboratory director of The Indiana University School of Medicine Pharmacogenomics Laboratory, a fee-for-service laboratory that offers clinical pharmacogenetic testing. S.C.S. and I.N. have a patent (US 20200168298A1) for the Aldy genotyping platform. All other authors declare no conflicts of interest.
Portions of this article were presented as an abstract at the 2022 American College of Medical Genetics and Genomics Annual Clinical Genetics Meeting, March 22–26, 2022, Nashville, TN.
Supplemental material for this article can be found at http://doi.org/10.1016/j.jmoldx.2022.03.008.
Supplemental Data
References
- 1.Ji Y., Si Y., McMillin G.A., Lyon E. Clinical pharmacogenomics testing in the era of next generation sequencing: challenges and opportunities for precision medicine. Expert Rev Mol Diagn. 2018;18:411–421. doi: 10.1080/14737159.2018.1461561. [DOI] [PubMed] [Google Scholar]
- 2.Tayeh M.K., Gaedigk A., Goetz M.P., Klein T.E., Lyon E., McMillin G.A., Rentas S., Shinawi M., Pratt V.M., Scott S.A., ACMG Laboratory Quality Assurance Committee Clinical pharmacogenomic testing and reporting: A technical standard of the American College of Medical Genetics and Genomics (ACMG) Genet Med. 2022;24:759–768. doi: 10.1016/j.gim.2021.12.009. [DOI] [PubMed] [Google Scholar]
- 3.Chen X., Shen F., Gonzaludo N., Malhotra A., Rogert C., Taft R.J., Bentley D.R., Eberle M.A. Cyrius: accurate CYP2D6 genotyping using whole-genome sequencing data. Pharmacogenomics J. 2021;21:251–261. doi: 10.1038/s41397-020-00205-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Lee S.B., Wheeler M.M., Patterson K., McGee S., Dalton R., Woodahl E.L., Gaedigk A., Thummel K.E., Nickerson D.A. Stargazer: a software tool for calling star alleles from next-generation sequencing data using CYP2D6 as a model. Genet Med. 2019;21:361–372. doi: 10.1038/s41436-018-0054-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Twist G.P., Gaedigk A., Miller N.A., Farrow E.G., Willig L.K., Dinwiddie D.L., Petrikin J.E., Soden S.E., Herd S., Gibson M., Cakici J.A., Riffel A.K., Leeder J.S., Dinakarpandian D., Kingsmore S.F. Constellation: a tool for rapid, automated phenotype assignment of a highly polymorphic pharmacogene, CYP2D6, from whole-genome sequences. NPJ Genom Med. 2016;1:15007. doi: 10.1038/npjgenmed.2015.7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Twesigomwe D., Drögemöller B.I., Wright G.E.B., Siddiqui A., da Rocha J., Lombard Z., Hazelhurst S. StellarPGx: a nextflow pipeline for calling star alleles in cytochrome P450 genes. Clin Pharmacol Ther. 2021;110:741–749. doi: 10.1002/cpt.2173. [DOI] [PubMed] [Google Scholar]
- 7.Klanderman B.J., Koch C., Machini K., Parpattedar S.S., Bandyadka S., Lin C.F., Hynes E., Lebo M.S., Amr S.S. Automated pharmacogenomic reports for clinical genome sequencing. J Mol Diagn. 2022;24:205–218. doi: 10.1016/j.jmoldx.2021.12.001. [DOI] [PubMed] [Google Scholar]
- 8.Numanagić I., Malikić S., Ford M., Qin X., Toji L., Radovich M., Skaar T.C., Pratt V.M., Berger B., Scherer S., Sahinalp S.C. Allelic decomposition and exact genotyping of highly polymorphic and structurally variant genes. Nat Commun. 2018;9:828. doi: 10.1038/s41467-018-03273-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Jones S., Anagnostou V., Lytle K., Parpart-Li S., Nesselbush M., Riley D.R., Shukla M., Chesnick B., Kadan M., Papp E., Galens K.G., Murphy D., Zhang T., Kann L., Sausen M., Angiuoli S.V., Diaz L.A., Jr., Velculescu V.E. Personalized genomic analyses for cancer mutation discovery and interpretation. Sci Transl Med. 2015;7:283ra53. doi: 10.1126/scitranslmed.aaa7161. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Kato S., Kim K.H., Lim H.J., Boichard A., Nikanjam M., Weihe E., Kuo D.J., Eskander R.N., Goodman A., Galanina N., Fanta P.T., Schwab R.B., Shatsky R., Plaxe S.C., Sharabi A., Stites E., Adashek J.J., Okamura R., Lee S., Lippman S.M., Sicklick J.K., Kurzrock R. Real-world data from a molecular tumor board demonstrates improved outcomes with a precision N-of-one strategy. Nat Commun. 2020;11:4965. doi: 10.1038/s41467-020-18613-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Amstutz U., Henricks L.M., Offer S.M., Barbarino J., Schellens J.H.M., Swen J.J., Klein T.E., McLeod H.L., Caudle K.E., Diasio R.B., Schwab M. Clinical Pharmacogenetics Implementation Consortium (CPIC) guideline for dihydropyrimidine dehydrogenase genotype and fluoropyrimidine dosing: 2017 update. Clin Pharmacol Ther. 2018;103:210–216. doi: 10.1002/cpt.911. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Relling M.V., Schwab M., Whirl-Carrillo M., Suarez-Kurtz G., Pui C.H., Stein C.M., Moyer A.M., Evans W.E., Klein T.E., Antillon-Klussmann F.G., Caudle K.E., Kato M., Yeoh A.E.J., Schmiegelow K., Yang J.J. Clinical Pharmacogenetics Implementation Consortium guideline for thiopurine dosing based on TPMT and NUDT15 genotypes: 2018 update. Clin Pharmacol Ther. 2019;105:1095–1105. doi: 10.1002/cpt.1304. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Crews K.R., Monte A.A., Huddart R., Caudle K.E., Kharasch E.D., Gaedigk A., Dunnenberger H.M., Leeder J.S., Callaghan J.T., Samer C.F., Klein T.E., Haidar C.E., Van Driest S.L., Ruano G., Sangkuhl K., Cavallari L.H., Müller D.J., Prows C.A., Nagy M., Somogyi A.A., Skaar T.C. Clinical Pharmacogenetics Implementation Consortium guideline for CYP2D6, OPRM1, and COMT genotypes and select opioid therapy. Clin Pharmacol Ther. 2021;110:888–896. doi: 10.1002/cpt.2149. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Hicks J.K., Bishop J.R., Sangkuhl K., Müller D.J., Ji Y., Leckband S.G., Leeder J.S., Graham R.L., Chiulli D.L., LLerena A., Skaar T.C., Scott S.A., Stingl J.C., Klein T.E., Caudle K.E., Gaedigk A. Clinical Pharmacogenetics Implementation Consortium (CPIC) guideline for CYP2D6 and CYP2C19 genotypes and dosing of selective serotonin reuptake inhibitors. Clin Pharmacol Ther. 2015;98:127–134. doi: 10.1002/cpt.147. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Cousin M.A., Matey E.T., Blackburn P.R., Boczek N.J., McAllister T.M., Kruisselbrink T.M., Babovic-Vuksanovic D., Lazaridis K.N., Klee E.W. Pharmacogenomic findings from clinical whole exome sequencing of diagnostic odyssey patients. Mol Genet Genomic Med. 2017;5:269–279. doi: 10.1002/mgg3.283. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Hutchcraft M.L., Lin N., Zhang S., Sears C., Zacholski K., Belcher E.A., Durbin E.B., Villano J.L., Cavnar M.J., Arnold S.M., Ueland F.R., Kolesar J.M. Real-world evaluation of universal germline screening for cancer treatment-relevant pharmacogenes. Cancers (Basel) 2021;13:4524. doi: 10.3390/cancers13184524. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Shugg T., Ly R.C., Rowe E.J., Philips S., Hyder M.A., Radovich M., Rosenman M.B., Pratt V.M., Callaghan J.T., Desta Z., Schneider B.P., Skaar T.C. Clinical opportunities for germline pharmacogenetics and management of drug-drug interactions in patients with advanced solid cancers. JCO Precis Oncol. 2022;6:e2100312. doi: 10.1200/PO.21.00312. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.van der Lee M., Allard W.G., Bollen S., Santen G.W.E., Ruivenkamp C.A.L., Hoffer M.J.V., Kriek M., Guchelaar H.J., Anvar S.Y., Swen J.J. Repurposing of diagnostic whole exome sequencing data of 1,583 individuals for clinical pharmacogenetics. Clin Pharmacol Ther. 2020;107:617–627. doi: 10.1002/cpt.1665. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Twesigomwe D., Wright G.E.B., Drögemöller B.I., da Rocha J., Lombard Z., Hazelhurst S. A systematic comparison of pharmacogene star allele calling bioinformatics algorithms: a focus on CYP2D6 genotyping. NPJ Genom Med. 2020;5:30. doi: 10.1038/s41525-020-0135-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.White T., Szelinger S., LoBello J., King A., Aldrich J., Garinger N., Halbert M., Richholt R.F., Mastrian S.D., Babb C., Ozols A.A., Goodman L.J., Basu G.D., Royce T. Analytic validation and clinical utilization of the comprehensive genomic profiling test, GEM ExTra(®) Oncotarget. 2021;12:726–739. doi: 10.18632/oncotarget.27945. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Numanagić I., Malikić S., Pratt V.M., Skaar T.C., Flockhart D.A., Sahinalp S.C. Cypiripi: exact genotyping of CYP2D6 using high-throughput sequencing data. Bioinformatics. 2015;31:i27–i34. doi: 10.1093/bioinformatics/btv232. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Gaedigk A., Boone E.C., Scherer S.E., Lee S.B., Numanagić I., Sahinalp C., Smith J.D., McGee S., Radhakrishnan A., Qin X., Wang W.Y., Farrow E.G., Gonzaludo N., Halpern A.L., Nickerson D.A., Miller N.A., Pratt V.M., Kalman L.V. CYP2C8, CYP2C9, and CYP2C19 characterization using next-generation sequencing and haplotype analysis: A GeT-RM collaborative project. J Mol Diagn. 2022;24:337–350. doi: 10.1016/j.jmoldx.2021.12.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Li H., Handsaker B., Wysoker A., Fennell T., Ruan J., Homer N., Marth G., Abecasis G., Durbin R., 1000 Genome Project Data Processing Subgroup The Sequence Alignment/Map format and SAMtools. Bioinformatics. 2009;25:2078–2079. doi: 10.1093/bioinformatics/btp352. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Belkadi A., Bolze A., Itan Y., Cobat A., Vincent Q.B., Antipenko A., Shang L., Boisson B., Casanova J.L., Abel L. Whole-genome sequencing is more powerful than whole-exome sequencing for detecting exome variants. Proc Natl Acad Sci U S A. 2015;112:5473–5478. doi: 10.1073/pnas.1418631112. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Lopes J.L., Harris K., Karow M.B., Peterson S.E., Kluge M.L., Kotzer K.E., Lopes G.S., Larson N.B., Bielinski S.J., Scherer S.E., Wang L., Weinshilboum R.M., Black J.L., 3rd, Moyer A.M. Targeted genotyping in clinical pharmacogenomics: what is missing? J Mol Diagn. 2022;24:253–261. doi: 10.1016/j.jmoldx.2021.11.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Nichols D., Arnold S., Weiss H.L., Wu J., Durbin E.B., Miller R., Kolesar J. Pharmacogenomic potential in advanced cancer patients. Am J Health Syst Pharm. 2019;76:415–423. doi: 10.1093/ajhp/zxy079. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Pratt V.M., Cavallari L.H., Del Tredici A.L., Gaedigk A., Hachad H., Ji Y., Kalman L.V., Ly R.C., Moyer A.M., Scott S.A., van Schaik R.H.N., Whirl-Carrillo M., Weck K.E. Recommendations for clinical CYP2D6 genotyping allele selection: a joint consensus recommendation of the Association for Molecular Pathology, College of American Pathologists, Dutch Pharmacogenetics Working Group of the Royal Dutch Pharmacists Association, and the European Society for Pharmacogenomics and Personalized Therapy. J Mol Diagn. 2021;23:1047–1064. doi: 10.1016/j.jmoldx.2021.05.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Gaedigk A., Twist G.P., Leeder J.S. CYP2D6, SULT1A1 and UGT2B17 copy number variation: quantitative detection by multiplex PCR. Pharmacogenomics. 2012;13:91–111. doi: 10.2217/pgs.11.135. [DOI] [PubMed] [Google Scholar]
- 29.Ramamoorthy A., Flockhart D.A., Hosono N., Kubo M., Nakamura Y., Skaar T.C. Differential quantification of CYP2D6 gene copy number by four different quantitative real-time PCR assays. Pharmacogenet Genomics. 2010;20:451–454. doi: 10.1097/FPC.0b013e32833a1083. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Rehder C., Bean L.J.H., Bick D., Chao E., Chung W., Das S., O'Daniel J., Rehm H., Shashi V., Vincent L.M. Next-generation sequencing for constitutional variants in the clinical laboratory, 2021 revision: a technical standard of the American College of Medical Genetics and Genomics (ACMG) Genet Med. 2021;23:1399–1415. doi: 10.1038/s41436-021-01139-4. [DOI] [PubMed] [Google Scholar]
- 31.Gaedigk A., Casey S.T., Whirl-Carrillo M., Miller N.A., Klein T.E. Pharmacogene Variation Consortium: a global resource and repository for pharmacogene variation. Clin Pharmacol Ther. 2021;110:542–545. doi: 10.1002/cpt.2321. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.