176
Z. Liu et al.
9.6 Closing Remarks
The critical assumption in using animal models was that the findings of the animal model would correlate to how the compound would behave in humans. Recent
researchers have shown this correlation to be poor, leading researchers to determine
if in vitro cell-based assays in combination with in silico approaches could enhance
3Rs principles. These researchers generated many novel techniques, and non-animal
testing means to tackle the issue. Toxicogenomics (TGx) is one of these techniques
that shows excellent promise to forecast toxicity in drug compounds while meeting
the 3Rs goal. Toward studying how TGx could be used in replacing animal models,
we developed a computational method called PRank, to address the transferability
among different TGx testing assays and promote the in vitro TGx.
In our study, we used data culled from the TG-GATEs database. TG-GATEs is a
large database for toxicogenomics, but there are many compounds not in the database.
Other drug transcriptional databases with more compounds were recently generated
and could be utilized to further probe PRank’s potential for assay transferability.
The most promising of these databases is the LINCS database [54]. The LINCS
database, which is publicly available, expands significantly upon the Connectively
Map (CMap). LINCS consists of more than 20 k compounds, and transcriptomic profiles were generated across more than 400 different cell lines. The LINCS database
shows how genes, drugs, and diseases are associated with common gene-expression
signatures [55]. Rather than use the entire human genome, the LINCS database
uses 1000 genes as “landmark” genes as representative genes and uses these landmark genes to extrapolate to the whole human genome. The LINCS database can be
accessed at https://clue.io. With the LINCS data set, we can further apply our PRank
method to address another question such as repurposing the transcriptomic profiles
from immortalized cell lines for toxicity assessment. Furthermore, the transferability between the assays in some novel cell culture such as iPSC and traditional cell
cultures could also be assessed.
The use of the PRank system is not confined to the areas of toxicogenomics. These
studies show that the PRank computational methodology can be successfully applied
to other types of data sets. For example, the PRank can be used to high throughput
screening assays from the Tox21 project or ToxCast. The National Institute of Environmental Sciences (NIEHS), part of the National Institute of Health (NIH), has a
multitude of open data sets, such as the Environmental Genome Project, that PRank
can be utilized to explore the strength of relationships.
As animal models begin to fall out of favor among compound safety studies, new
systems must be ready to assess the potential toxicity of new compounds in humans.
Utilizing already existing and open toxicogenomic databases present an opportunity
to develop novel in vitro and in silico strategies to assess this potential toxicity.
The PRank computational tool is a promising approach to bridge the gap between
the available data and to gain insight into how new compounds may present their
toxicity to humans, and thus answer some of the most significant questions in the
toxicology field.
Z. Liu et al.
9.6 Closing Remarks
The critical assumption in using animal models was that the findings of the animal model would correlate to how the compound would behave in humans. Recent
researchers have shown this correlation to be poor, leading researchers to determine
if in vitro cell-based assays in combination with in silico approaches could enhance
3Rs principles. These researchers generated many novel techniques, and non-animal
testing means to tackle the issue. Toxicogenomics (TGx) is one of these techniques
that shows excellent promise to forecast toxicity in drug compounds while meeting
the 3Rs goal. Toward studying how TGx could be used in replacing animal models,
we developed a computational method called PRank, to address the transferability
among different TGx testing assays and promote the in vitro TGx.
In our study, we used data culled from the TG-GATEs database. TG-GATEs is a
large database for toxicogenomics, but there are many compounds not in the database.
Other drug transcriptional databases with more compounds were recently generated
and could be utilized to further probe PRank’s potential for assay transferability.
The most promising of these databases is the LINCS database [54]. The LINCS
database, which is publicly available, expands significantly upon the Connectively
Map (CMap). LINCS consists of more than 20 k compounds, and transcriptomic profiles were generated across more than 400 different cell lines. The LINCS database
shows how genes, drugs, and diseases are associated with common gene-expression
signatures [55]. Rather than use the entire human genome, the LINCS database
uses 1000 genes as “landmark” genes as representative genes and uses these landmark genes to extrapolate to the whole human genome. The LINCS database can be
accessed at https://clue.io. With the LINCS data set, we can further apply our PRank
method to address another question such as repurposing the transcriptomic profiles
from immortalized cell lines for toxicity assessment. Furthermore, the transferability between the assays in some novel cell culture such as iPSC and traditional cell
cultures could also be assessed.
The use of the PRank system is not confined to the areas of toxicogenomics. These
studies show that the PRank computational methodology can be successfully applied
to other types of data sets. For example, the PRank can be used to high throughput
screening assays from the Tox21 project or ToxCast. The National Institute of Environmental Sciences (NIEHS), part of the National Institute of Health (NIH), has a
multitude of open data sets, such as the Environmental Genome Project, that PRank
can be utilized to explore the strength of relationships.
As animal models begin to fall out of favor among compound safety studies, new
systems must be ready to assess the potential toxicity of new compounds in humans.
Utilizing already existing and open toxicogenomic databases present an opportunity
to develop novel in vitro and in silico strategies to assess this potential toxicity.
The PRank computational tool is a promising approach to bridge the gap between
the available data and to gain insight into how new compounds may present their
toxicity to humans, and thus answer some of the most significant questions in the
toxicology field.