computed. This can be done using the extrinsic spike-in molecules
as described above—the extent of variability in their expression
across cells can be used as an estimate of the null variance. This
information allows the expected technical variation to be modeled
across the whole dynamic range of expression, which forms the
basis of a statistical test to determine the set of genes that show
more variability in expression than would be expected by chance
[82, 94]. Recently, extrinsic spike-in molecules have been used to
further decompose technical variability into two terms that corre-
spond to sampling noise and heterogeneity in sequencing efficiency
across cells [151].2.4.4 Allele-Specific
Expression
scRNA-seq could in principle be used to study on the degree to
which the two alleles of each gene are regulated in a coordinated
manner. Like bulk RNA-seq studies [152], allele-specific expression
can be measured and used to determine the extent of allelic bias in
gene expression [150]. By exploring the degree of allele-specific
expression, stochastic transcription of each allele and the degree of
coordination of expression between alleles can be investigated. For
example, scRNA-seq has been used to study stochastic allelic
expression during early embryogenesis [150]. Specifically, using
first-generation intercrosses between two different inbred strains
of mice, the extent of stochastic allele-specific expression during
early embryogenesis has been quantified transcriptome-wide.
One concern with using scRNA-seq to study random mono-
allelic expression is that allelic dropout during library preparation
might lead to erroneous measurements of monoallelic expression.
Previous approaches have addressed this by splitting cell lysates into
two and then repeating the experiment to provide a background
estimate of allelic dropout [150]. However, this is an area in which
more work is required to develop computational methods that can
accurately model this feature of scRNA-seq library preparation,
such that accurate measures of allele-specific expression can be
obtained.3 Computational Approaches for Single-Cell Genomics
The field of single-cell genomics is developing rapidly and is gen-
erating increasingly new insights into complex biological systems.
In contrast to RNA-seq, DNA-seq is still costly; thus, to ensure
adequate and appropriate data are obtained to address the hypoth-
eses of the study, the balance between the number of cells
sequenced and the breadth of the genome of each cell that will be
queried needs to be taken into account when designing an experi-
ment. Broadly speaking, for complex eukaryotic genomes such as
the human genome, one can choose to query specific loci of interest358 Yungang Xu and Xiaobo Zhou