Exome Sequencing as a Discovery and Diagnostic Tool - Next Generation Sequencing Technologies in Medical Genetics - page 73

hyperphosphatasia mental retardation syndrome (Krawitz et al. 2010 ; Musunuru
et al. 2010 ; Johnson et al. 2010b ). In addition, WES has also been swiftly integrated
with homozygosity mapping to accelerate the investigation of recessive disorders in
consanguineous families (Walsh et al. 2010 ; Bolze et al. 2010 ).

8.4.3 Implementation of Family-Based Analysis

as a Discovery Tool

Tens of thousands of single nucleotide variants and short indels are detected in the
sequencing of the human exome and multiple fi ltering criteria are typically
employed for the identifi cation of causal variants. These fi lters include inheri-
tance models, frequency, mutation types (nonsynonymous variants, splice-site
disruptions, or coding indels), predictions, functional classifi cation, and literature
fi ndings, to mention a few. As shown in the previous sections, these fi lters have
proven effective in identifying candidate genes for several Mendelian disorders
and the fi ltering process, which is the essence of exome analysis, will be discussed
below.

8.4.3.1 Primary, Secondary, and Tertiary Variant Filtering

The general workfl ow of NGS variant detection analysis is summarized in three
major phases. The primary phase is the process of generating base calls combined
with quality control fi ltering, which is performed automatically after the image
acquisition (Illumina HiSeq2000/MiSeq, Roche/454 Genome Sequencer) on the
instrument-attached computer, and generates FASTQ fi les for the second step
(Fig. 8.1 ). The throughput depends on the instrument, typically producing 30-45,000
variants per exome on 100 paired-end cycles.
The secondary phase is the process of alignment to the human genome reference
sequence combined with quality fi ltering to remove common sequencing errors
(Fig. 8.1 ). Typically, 98 % of the high quality reads are expected to align to the
human genome reference sequence (Genome Reference Consortium, GRCh37).
Directly after the alignment, a process of determining variants by comparison of
consensus to genome reference is performed. This process of variant detection is
followed by an annotation step where gene features, such as functional predictions,
conservation, and others, are added to each variant. Example of commercial soft-
ware that perform thousands of variant annotations per hour is Alamut-HT and adds
the information from other databases such as dbSNP, OMIM, and HGMD and pro-
vides precomputed predictions for various prediction tools, like and ANNOVAR.
Examples of prediction tools that are publicly available are PANTHER (Protein
Analysis Through Evolutionary Relationships), SIFT (Sorting Intolerant From
Tolerant amino acid substitutions), and Polyphen-2.