Commercialization of HaploSeq as a Service (HaaS) for generating chromosome-span phased genome and exome sequence information - Commercialization of HaploSeq as a Service (HaaS) for generating chromosome-span phased genome
and exome sequence information
Arima Genomics
SPECIFIC AIMS
7. Project Summary/Abstract
Over 90% of Next-Generation Sequencing (NGS) is sequenced via Illumina short-read sequencers. This is
because of its cost-effectiveness and faster turn-around times. However, short-read sequencing technologies
lose critical contiguity information and are limited in assembling genomes de novo and reconstructing maternal
and paternal haplotypes of diploid genomes. Contiguity information is valuable for understanding the genetics
of human health and disease, and therefore critical for advancing precision and personalized medicine. Long-
read technologies (e.g. Pacific Biosciences) only reach megabase-scale chromosomal contiguity, but is 5-7X
more expensive than Illumina short-read, limiting its use. Recent advances in DNA preparation can preserve
long-range information that is compatible with Illumina short-read sequencing. These “synthetic” long-reads (or
SLR) methods can improve short-read technologies with, long-range contiguity and short-read economy.
However, the maximal SLR contiguity is only 1-5% the contiguity a 100Mb average human chromosome. To
construct multi-megabase contiguity, SLR methods require genomic DNA (gDNA) fragments >100-150Kb, but
obtaining long gDNA fragments is challenging and this limits current SLR methods.
Arima Genomics has optimized HiC technology, an SLR-based DNA protocol, to establish Arima HiC (A-HiC)
that preserves chromosome-span contiguity for de novo assembly, haplotype phasing and metagenomics, the
libraries of which can be sequenced via Illumina short-read instruments. A-HiC, rather than using purified
gDNA, leverages the long-contiguity information preserved naturally in the 3-dimensional (3D) organization of
genomes in cells. Indeed, 3D information is not only long-range but in fact full chromosome-range information.
A-HiC optimizes multiple features of HiC – while HiC is laborious, time-consuming 3-day procedure, costly
protocol, A-HiC is easy to perform, generates consistent quality libraries, >70% less cost and is only a 6-hour
protocol, and is compatible to standard library preps such as KAPA Hyper preps – together, these properties of
A-HiC make them automatable. After success with manual automation via 96-well plates, we propose to use
liquid handler (in partnership with Agilent) to automate A-HiC, and furthermore, we aim to make A-HiC robust
to wide-range of sample types (cells, tissues, blood, human and non-human) to serve diverse customers via
the automated service platform, referred to as HaaS. In addition to optimizing experimental A-HIC, we develop
several algorithms to generate chromosome-span phase information of genomes and exomes, which we will
publish as open source software (OSS). We also leverage existing OSS for other HiC-based apps, and
together we will automate software for all HiC-enabled apps in supercomputing infrastructure to enable quick
turn around time for our diverse customers. Together, HaaS architecture has automated experimental (A-HiC)
and computational aspects (OSS for HiC-based apps).
Many commercial players (e.g. Novogene) provide sequencing services. Arima's HaaS provides chromosomal
contiguity and thus is differentiated from the traditional (category-1) service providers such as Novogene who
provide fragmented contiguity based on short-read or long-read methods. Indeed, we propose to collaborate
with Novogene. We are also in communication with PacBio for a potential collaboration and marketing
agreement. On the other hand, we compete directly with category-2 players who offer HiC services, specifically
for de novo assembly application. Nonetheless HiC services from other category-2 companies suffer significant
limitations – (1) high prices (>$10,000 while Arima prices at <$5,000 for large genomes and <$3,000 for small
genomes), (2) low quality data via usage of traditional HiC (while Arima uses optimized A-HiC), (3) non-
automatable traditional HiC (while Arima uses fully automatable A-HiC), and in addition, we have developed
specialized phasing algorithms to garner wide customer base. To date, Arima has barely marketed our
services, but word-of-mouth has garnered new customers for Arima and attracted repeat business, which
reflects the quality of Arima's services and demonstrates significant market demand for Arima – setting the
stage for rapid growth to be supported by more deliberate traditional marketing of our proposed HaaS business
model.
In this proposal, we develop and benchmark HaaS via collaboration with Key Opinion leaders across multiple
sample types (blood, cells, tissues, human and non-human samples) for several HiC-based apps.
