Targeted next-generation sequencing panels enable interrogation of multiple genes across many samples to more deeply understand human genetic disease. However, finding all relevant genes, developing robust, high performing multiplex panels, and implementing scalable, reproducible and accurate analysis pipelines is challenging. We present a coordinated suite of tools to facilitate genetic disease research. First, we developed the Content Selection Engine which organizes human diseases hierarchically, and links all diseases to a set of associated genes; and the Gene Scoring Algorithm which ranks genes by clinical relevance. We developed optimized assays for the most studied 1000 disease research genes, and we are in the process of developing optimized assays for a further 4000 genes. An interactive web interface allows scientists to select any disease of interest, display all associated genes, select any genes, and add additional genes, for any number of diseases. Empirical coverage for each gene can be visualized in IGV. A custom Ampliseq gene panel can be built using the optimized assays from all the selected genes. Optimized gene panels can be developed narrowly targeted to specific diseases, or larger gene panels can be developed for broader phenotypes. Disease categories include early onset neonatal phenotypes such as metabolic disorders, Severe Combined Immunodeficiency (SCID), heme disorders; and late onset disorders such as cancer predisposition and cardiovascular disorders.

1. Thermo Fisher Scientific • 5781 Van Allen Way • Carlsbad, CA 92008 • thermofisher.com
Fiona Hyland, Mani Manivannan, Brijesh Krishnaswami, Yun Zhu, Yuan Tian, Cristina Van Loy, Adam Broomer, Tanya Biorac, Dominique DeWolf, Yutao Fu, Mark Andersen, Francisco
Hernandez Guzman, Arvind Kothandaraman, Steve Roman. Thermo Fisher Scientific, 180 Oyster Point Blvd, South San Francisco, CA , USA and Carlsbad, CA, USA.
Developing Custom Next-Generation Sequencing Panels using Pre-Optimized Assays: an Integrated approach from
Disease Research Area to Functionally Annotated Variants with Ion Ampliseq On-Demand Panels.
INTRODUCTION
Targeted next-generation sequencing panels enable interrogation of multiple genes across many samples to more deeply understand
human genetic disease. However, finding all relevant genes, developing robust, high performing multiplex panels, and implementing
scalable, reproducible and accurate analysis pipelines is challenging. We present a coordinated suite of tools to facilitate genetic disease
research. First, we developed the Content Selection Engine which organizes human diseases hierarchically, and links all diseases to a set
of associated genes; and the Gene Scoring Algorithm which ranks genes by clinical relevance. We developed optimized assays for the
most studied 1000 disease research genes, and we are in the process of developing optimized assays for a further 4000 genes.
An interactive web interface allows scientists to select any disease of interest, display all associated genes, select any genes, and add
additional genes, for any number of diseases. Empirical coverage for each gene can be visualized in IGV. A custom Ampliseq gene panel
can be built using the optimized assays from all the selected genes. Optimized gene panels can be developed narrowly targeted to
specific diseases, or larger gene panels can be developed for broader phenotypes. Disease categories include early onset neonatal
phenotypes such as metabolic disorders, Severe Combined Immunodeficiency (SCID), heme disorders; and late onset disorders such as
cancer predisposition and cardiovascular disorders.
The panel information is integrated into a seamless end to end analysis. Single sample, Trio, and Tumor-Normal pipelines are available.
Pipelines include mapping and variant calling, followed by functional annotation, protein prediction, and detailed annotation of variants and
genes based on their presence in public databases. Filtering and reporting completes the analysis.
Run
Sequence
Prepare
Library
Sample
Input
Analyze
Prepare
Template
Oncomine™
Immune Response
Research Assay
(Manual or
automated library
preparation)
FFPE derived total
RNA
Analysis solution on
Torrent SuiteTM
Ion S5™
System
(520 or 530)
or
Ion PGM™ System
(318)
Ion Chef™ System
or
Ion OneTouch™ 2
System
Content Selection Engine
The synthesis of MeSH disease hierarchies with DisGeNET gene-disease relations forms our Content Selection Engine. MeSH provides a
mathematical graph indicating disease parent/child relationships; e.g., “autoimmune disease” is a parent disease of “rheumatoid arthritis”,
while “juvenile rheumatoid arthritis” is a child disease of “rheumatoid arthritis”.
The nodes on the graph, the diseases themselves, are related to genes through a score provided by DisGeNET for each gene-disease pair.
Taken together, the disease hierarchy combined with the gene-disease scores provide a way to look up genes involved with any disease or
group of diseases at any level in the disease hierarchy. The Gene Scoring Algorithm (GSA) summarizes child diseases’ gene contributions
to a selected parent disease in the hierarchy.
The database contains over X associations between Y genes and Z diseases, both Mendelian and complex.
METHODS
RESULTS
CONCLUSIONS
Panel based Uniformity on dried blood spots,
frozen samples (examples from 2 panels)
Robust, high performing Ampliseq On-Demand Gene Panels
Ampliseq On Demand assays are designed to comprehensively cover exons. While the minimum
exon padding is 5 bases, typically genes have amplicons extending 50 bases or more past the edge
of the exon. Number of amplicons per gene ranges between 1 and 303.
The optimizaton process resulted in an optimized set of gene assays which have high performance:
most of the genes have 100% in silico coverage and 100% uniformity. Uniformity is defined as the %
of bases on target having read depth ≥ 0.2X of the mean read depth. Uniformity is typically defined
per panel, and can also be measured per gene.
Disease Gene Scoring
Rank-Weighted Sum Score (RWSS) is an unbiased gene scoring method
that accounts for both the strength and number of gene-disease pairs.
The GSA module takes the DisGeNET scores as input and uses RWSS method to
prioritize genes for a specific disease of interest.
In conjunction with the DAD module, the GSA module is able to produce a list of
ranked genes for one or more diseases at any level of the disease hierarchy.
The DisGeNET score was developed to rank the gene-disease associations
according to their level of evidence. DisGeNET gene-disease association score
takes into account the number and type of sources (level of curation, model
organisms), and the number of publications supporting the association. The score
ranges from 0 to 1 . Per DisGeNet, a score of 0.1 corresponds to an average of
about 3 sources of evidence. A score of 0.25 corresponds to between 4 and 5
sources of evidence for single gene-disease assocation.
Optimized Ampliseq On Demand Genes
We developed a “Circle of Optimization” process for designing,
measuring, refining and redesigning gene Assay designs. We
developed a custom assay design pipeline to incorporate all
available information. Gene designs were manufactured , their
performance was measured as part of development pools, and
then 10 representative disease panels of sizes ranging from 17 to
300 genes (322 to 8320 amplicons) were created and their
performance was measured.
All genes are created as 2-pool designs, with a maximum amplicon
length of 275 bases.
Seamless Design of Ion Ampliseq On-Demand Gene Panels based on Diseases or Gene List
Selecting genes based on browsing Disease hierarchy
We developed an interactive web site running on AWS to facilitate seamless, fast, comprehensive design of gene panels. This is accessible at
Ampliseq.com . A custom Ampliseq On-Demand gene panel can be built using the optimized assays from any subset or all of the selected genes.
Optimized gene panels can be developed narrowly targeted to specific diseases, or larger gene panels can be developed for broader phenotypes.
Browsing through the disease hierarchy reveals the number of genes associated with each disease at every level of the hierarchy. Any number of
diseases at any combination of levels can be selected, leading to a list of genes associated with each of the diseases, ordered by score. Genes
available pre-optimized on-demand can be selected or unselected to create an Ion Ampliseq On-Demand Panel. Additional genes can be added
manually.
Designing panels based on Gene List
Alternatively, genes can be selected without going through the disease hierarchy by searching for genes of interest, or by uploading a gene list.
Panel Sizes offered
Panels can be ordered in two size ranges: 1-50 genes or 51-300 genes. The minimum number of amplicons is 24, maximum is 9000 amplicons.
Adding Spike-In Panels
Genes of interest that are not yet available on-demand can be ordered in a separate panel that can be spiked in to the Ion Ampliseq On-Demand
panel. The spike-in panel will be custom manufactured and is available in large reaction sizes.
Observed Performance by base, amplicon, gene, with IGV
All of the genes on-demand have empirical performance data from assay development and verification. This performance information is available on
Ampliseq.com. IGV enables visualization of the genomic positions of exons, the amplicons that cover them, and the empirical observations of
coverage of each base and each amplicon for all genes in every custom Ion Ampliseq On-Demand Panel. This enables confidence that amplicons
are designed to cover each important base and that observed performance is good for each base.
Analysis: Variant Calling through Annotation, Functional Impact, to Prioritization
Ion Reporter software provides a seamless end to end analysis for Ampliseq On-Demand panels, including mapping, variant calling, and
CNV detection. Single sample, Trio, and Tumor-Normal pipelines are available. Pipelines include mapping and variant calling, followed by
functional annotation, protein prediction, and detailed annotation of variants and genes based on their presence in public databases with
link-outs to external sites. The software enables the essential steps of annotation, filtering, and prioritization, and finally reporting.
References, Additional Information
See Poster P14.069A ‘A novel method for building custom AmpliSeq™ panels using optimized PCR primers.’ Steve Roman et al.
DisGeNET and Gene Scoring
http://www.disgenet.org/web/DisGeNET/menu/dbinfo#score
Level 1 Diseases Examples of Genes Associated with each Disease
# of total
Genes
# of Ampliseq
On Demand
Genes
Cardiovascular Diseases NPPA, ZFPM2, TGFB1, TGFBR1, TTN, STAT3, FLT1, TMEM43, HLA-A, MYLK, IGF2, AVPR1A, NPHS1, MYL3, HOXA1, PDE8B… 1375 319
Congenital, Hereditary, and
Neonatal Diseases and
Abnormalities
MECP2, TP63, VHL, LHCGR, XPC, RASA1, GNAS, KCNH2, MFN2, ACADVL, HBA1, DKC1, IVD, PITX2, CYP1B1, RHO, SMAD3, KRAS,
RDH12, TFAP2A, HMOX1, ABCB1, GCLC, CBFB, DNAH5, SLC40A1, APP, ETFDH, PSEN1, SYNE2, CCDC40, GNA11, MYL3, BARD1... 1743 641
Digestive System Disorders
APC, MTHFR, SLC37A4, ABCB4, FH, F2, ARID1A, NOTCH2, C3, PRKDC, KCNQ1, IL7R, PEX5, POLG, PALB2, SLIT2, GLI3, FBN2,
FOXA1, IRX5, CARD11... 1660 387
Disorder of eye KIT, OPTN, SOD2, NOS3, FKRP, TGFBR2, GATA6… 473 128
Endocrine System Diseases
TP53, CYP11B1, COL3A1, NR0B1, RECQL4, CHEK2, PIK3CA, EP300, BMP2, BAP1, SERPINA1, RYR2, BARD1, DPYD, SHOC2, APOE,
ACTG1, MAML3… 982 349
Female Urogenital Diseases
and Pregnancy Complications
FSHR, SCNN1B, HSD17B3, PPARG, HLA-B, MMP2, NR3C1, DICER1, SLC46A1, CSF1R, IFT122, SERPINA1, POLG, FANCA, VCAN,
INF2, FAT4, KDM6A, CSMD3… 2085 363
Hemic and Lymphatic
Diseases
FLT3, TBX1, TP53, TGFB1, ATM, MTR, PPARG, CFI, IGF2, AKT1, CTNNB1, RET, RP1… 659 263
Immune System Diseases
VWF, WT1, PTPN11, TERT, FLT4, CHEK2, LHCGR, CASP8, HSD3B2, MAPK1, MID1, TLX1, EPHX2, TFR2, TRAF1, CALR, NPPA, FGFR4,
GH1, DMD, NKX2-5, DNAH5, IGSF1… 1620 439
Male Urogenital Diseases ESR1, CHEK2, APC, POLG, HSD17B3, NFE2L2… 1741 117
Mental disorders
PAX6, NF1, COL11A2, PAX3, GNAQ, ZEB2, KCNH2, CYP11B1, PTGS2, FRAS1, CYP11B2, KCNE2, GH1, DLD, HFE, APC, EYS, PSEN2,
MTRR, ESR1, HAND2, FADD, CFI, FAS, APOB, ETV6, STAT1, PRDM1, TLX1, DEFA5, NKX2-6… 1355 572
Musculoskeletal Diseases
COL1A1, NOS3, POMC, RET, MMP2, GHR, VDR, TFAP2B, NKX2-5, CACNA1A, RAF1, CHEK2, NRXN1, FLG, PTGIS, HP, MLC1,
EDARADD, KRAS, HLA-A, NTF3, CYP2C9, LTBP2, SYNE1, BRIP1, CASP1, BAP1, BARD1, PROKR2, BMP6, HBA1, PEX13, IRS1, MYO6,
RUNX1, MAMLD1, SPG7, SLC40A1, MYL3, KRT6B, RBM20...
1117 768
Neoplasms
PTEN, KRAS, PTCH1, CREBBP, XPC, CHEK2, WFS1, HBA1, GATA2, TSHR, ARID1A, SMO, CDKN2B, CP, CSF1R, PIK3CG, MUC5B,
FOLR1, IFT122, SETD2, ACTN4, CALM2, SLIT2, SF3B1, DSG2, SLC29A3, TPMT, FSHR, SGCB, MAGEC1, PARP4, RAD21, MYOT, TPR,
BCLAF1, HSD17B10…
2851 674
Nervous System Disorder
KCNQ1, RET, SLC26A2, EYA1, SOD1, TP53, SOD2, COL3A1, CYP1B1, BLM, EDA, HSD17B3, KCNE1, MAP2K2, SOX2, GATA6, LRRK2,
DBT, IRX5, AVPR1A, CYP11A1, SETBP1, CASP1, PMS2, SLIT2, CACNA1S, MAGEC1, CIC, CBFB, ABCG2, TG, CDH1, KRT6A, MUTYH,
TAT, BTD, CHD4…
2186 709
Nutritional and Metabolic
Diseases
FGFR3, NOS3, TGFB1, SLC46A1, OCA2, ACADVL, SERPINA1, RAG2, DLD, GALE, IKBKG, MUT, EPHX2, FTL, PON2, NOD2, AVPR1A,
ITIH4, MYLK, KCNJ18… 1317 368
Otorhinolaryngologic Diseases DSPP, PIK3CA, TNF, COL2A1, TJP2, KRT6B… 195 111
Respiratory Tract Diseases
NOS3, ZFPM2, ESR1, TGFBR2, LPIN2, NFE2L2, TNC, APOC3, ANK2, ATF3, IGF2R, MYLK, MECP2, ACTN2, COL1A2, PCSK9, BSN,
PTCH1, FOXC2, ACTG1, DYX1C1, SYNE2… 804 418
Skin and Connective Tissue
Diseases
MECP2, WT1, COL5A1, MPZ, SOX9, NPPA, FGD1, CHD7, LYST, TLR4, HNF1B, RHO, KCNE2, MAP2K1, FH, TAZ, PRKCSH, RDH12,
MITF, DPYD, IFT140, LRRK2, ATF3, CBFB, IFT88, DES, SYNE2, SRGAP3, SMARCB1, SMAD2, ESPN, BCAT2, CSF1R, MAP3K1, FGFR4,
HABP2, SGCB, STXBP1, SRPX2, DRC1...
1547 767
Stomatognathic Diseases HLA-B, PCSK5, FAT1, KRT6B… 298 72
New Born Screening
HBB, CYP21A2, MMACHC, MMADHC, KCNQ1, HPD, GJB2, MTHFR, TAZ, PCCB, PCCA, GCH1, QDPR, MCCC1, MAT1A, CBS, PTPN11,
TAT, GJB3, MTRR, TPO, TSHR, DUOX2, FAH, GNMT, GJB6, AHCY, MTR, TG, G6PD, IDUA, HBA1, SLC25A13, PAX8, HBA2, SLC5A5,
CPT2, PTS, PCBD1, DECR1, MCCC2, TSHB, ACADSB, HADH, MMAA, OPA3, ACADVL, AUH, DNAJC19, HMGCL, HSD17B10, MCEE,
MMAB, MUT, CFTR, JAG1, FBN1, ZFPM2, ARSB, GAA, ABCD1, IDS, KCNH2, NKX2-5, KCNJ2, LAMP2, ACADM, GUSB, BTD, ASS1,
DNAI1, SLC22A5, ACADS,,
276 276
ACMG APC, MYH11, ACTA2, MYLK, TMEM43, DSP, PKP2, DSG2, DSC2, BRCA1, BRCA2, SCN5A,, WT1 … 59 59
0
50
100
150
200
250
300
350
400
10
20
30
50
75
100
125
150
200
303
#ofGenes
Number of Amplicons per Gene
0
1000
2000
3000
4000
5000
6000
7000
5
10
15
20
30
40
50
75
100
150
200
225
#Amplicons
Bases overlapping between adjacent
amplicon inserts
0
2000
4000
6000
8000
10000
12000
14000
0 10 20 50 100 150 200 250
NumberofExons
Exon Padding (bases from Amplicon edge to
Exon)
Disease Categories and Associated Genes
Genes associated with the Disease Research areas at Level 1 of the hierarchy. We have >1000 Ampliseq On-Demand genes as of May 2017
For Research Use Only. Not for use in diagnostic procedures. © 2017 Thermo Fisher Scientific Inc. All rights reserved.
All trademarks are the property of Thermo Fisher Scientific and its subsidiaries unless otherwise specified.
Panel Based Uniformity: uniformities of libraries prepared from matched
frozen (darker bars) and dried blood samples (lighter bars) for two
example panels; a hematology panel with 136 genes, and a Rheumatoid
Arthritis panel with 30 genes. Sample input was 10 ng. Error bars
represent ± 1 SD
Sequencing the Panel with the Ion S5 System
Ampliseq libraries are run on Ion sequencing instruments, including
the Ion S5 System. Sequence from DNA library to data is available in
as little as 24 hours with only 45 minutes of total hands-on time.
Custom
Assay
Design
Pipeline
Wet Lab
Testing,
Analysis,
Metrics
Amplicon
Performance
Database
Identify
Amplicons
that pass or
fail criteria
Select
Verified
Amplicons
We describe a full end-to-end solution for selecting genes associated with disease research areas of interest : creating a panel for
next-generation sequencing with these genes with amplicons that have already passed through optimization and performance
verification : sequencing samples with this panel using as little as 10ng of input DNA : and performing automated, straightforward
data analysis of the DNA sequence through to SNP detection, annotation, filtering and prioritization.
Gene
Design
Complete