sequencing analyzed 21 breast tumors and concluded, via the study of mutations that are
present in only a small fraction of the sequencing reads, that all samples had more subclonal
than clonal mutations and authors were able to reconstruct phylogenetic trees for some of
the tumors [27]. Since then, many other studies have been published where similar analyses
have informed our understanding of tumor evolution, selection dynamics, and mutation
cooperation [26]. This approach demonstrates the power of careful analysis of sequencing
reads and reveals the amount of information that can be inferred from these experiments.
Bulk transcriptome sequencing can also be examined beyond the differential expression
analyses that are typically done with such data. A number of different methodologies have
been developed to perform “cell type deconvolution,” a methodology used to infer the
proportions of different cell types in a tissue sample using computational approaches based
on specific marker genes or expression signatures. Deconvolution methods quantitatively
estimate the fractions of individual cell types in a heterocellular tissue (such as the tumor
microenvironment) by considering the bulk transcriptome as the “convolution” of cellspecific signatures [28]. One of the most popular method is CIBERSORT [28], which is
able to estimate the immune cell fraction component of a tumor biopsy. This algorithm has
been used to identify immune infiltration signatures in distinct cancer types and their
relationship to survival patterns and other clinical characteristics [29]. Other similar
methods are MuSiC, deconvSeq, and SCDC [30]. The choice of method would depend
on the type of tissue being studied and the biological question being addressed.
Of course, the exploration of intra-tumor heterogeneity has been spectacularly boosted
by the development of single-cell sequencing technologies. Recent studies exploiting this
technology have been able to show that inter-tumor heterogeneity in cancer cells is much
larger than intra-tumor heterogeneity, whereas this is not the case for non-malignant cells,
and to dissect patterns of heterogeneity (such as cell cycle stage and hypoxia response)
from context-specific programs that determine tumor progression and drug response
[31]. These advances would have been impossible without the ability to discern cell
types and cell states within a tumor and have greatly informed our understanding of
tumor dynamics.
2.6
Sequencing in Cancer Treatment
Perhaps the area in which tumor sequencing has had a more tangible impact has been in
precision and personalized treatment design. Whole-genome and -exome sequencing of
large groups of tumors has made it possible to identify genomic and transcriptomic
subtypes in neoplasia from tissues such as breast, skin, and bladder, among others
[32]. These genomic subtypes are associated with different clinical presentations and
molecular characteristics, and can be targeted with different treatments. For example, in
melanoma, four genomic subtypes have been identified: Tumors that have BRAF mutated,
almost always at the V600 residue, which present at a younger age and can be targeted with
BRAF inhibitors such as vemurafenib, those that have a RAS gene mutated, characterized
28
C. Molina-Aguilar et al.
present in only a small fraction of the sequencing reads, that all samples had more subclonal
than clonal mutations and authors were able to reconstruct phylogenetic trees for some of
the tumors [27]. Since then, many other studies have been published where similar analyses
have informed our understanding of tumor evolution, selection dynamics, and mutation
cooperation [26]. This approach demonstrates the power of careful analysis of sequencing
reads and reveals the amount of information that can be inferred from these experiments.
Bulk transcriptome sequencing can also be examined beyond the differential expression
analyses that are typically done with such data. A number of different methodologies have
been developed to perform “cell type deconvolution,” a methodology used to infer the
proportions of different cell types in a tissue sample using computational approaches based
on specific marker genes or expression signatures. Deconvolution methods quantitatively
estimate the fractions of individual cell types in a heterocellular tissue (such as the tumor
microenvironment) by considering the bulk transcriptome as the “convolution” of cellspecific signatures [28]. One of the most popular method is CIBERSORT [28], which is
able to estimate the immune cell fraction component of a tumor biopsy. This algorithm has
been used to identify immune infiltration signatures in distinct cancer types and their
relationship to survival patterns and other clinical characteristics [29]. Other similar
methods are MuSiC, deconvSeq, and SCDC [30]. The choice of method would depend
on the type of tissue being studied and the biological question being addressed.
Of course, the exploration of intra-tumor heterogeneity has been spectacularly boosted
by the development of single-cell sequencing technologies. Recent studies exploiting this
technology have been able to show that inter-tumor heterogeneity in cancer cells is much
larger than intra-tumor heterogeneity, whereas this is not the case for non-malignant cells,
and to dissect patterns of heterogeneity (such as cell cycle stage and hypoxia response)
from context-specific programs that determine tumor progression and drug response
[31]. These advances would have been impossible without the ability to discern cell
types and cell states within a tumor and have greatly informed our understanding of
tumor dynamics.
2.6
Sequencing in Cancer Treatment
Perhaps the area in which tumor sequencing has had a more tangible impact has been in
precision and personalized treatment design. Whole-genome and -exome sequencing of
large groups of tumors has made it possible to identify genomic and transcriptomic
subtypes in neoplasia from tissues such as breast, skin, and bladder, among others
[32]. These genomic subtypes are associated with different clinical presentations and
molecular characteristics, and can be targeted with different treatments. For example, in
melanoma, four genomic subtypes have been identified: Tumors that have BRAF mutated,
almost always at the V600 residue, which present at a younger age and can be targeted with
BRAF inhibitors such as vemurafenib, those that have a RAS gene mutated, characterized
28
C. Molina-Aguilar et al.