8 An Overview of National Toxicology Program’s Toxicogenomic …
147
Fig. 8.3 A high-level overview of the main information items in the DrugMatrix data warehouse
and user interface. The major information domains used by the DrugMatrix GUI are shown in gold,
green, orange, and blue rectangles. This represents a highly simplified in that it represents only 13
of the 70+ tables in the warehouse
The DrugMatrix relational database is available for download (PostgreSql scripts;
see below) and will be accessible soon through a database client. A Readme will be
available through CEBS which will provide details on how to connect to DrugMatrix
PostgreSql Database.
The two primary data domains, GENE and COMPOUND, are connected via
annotation of chemogenomic-based secondary domains: ASSAY activities (a direct
measure of interaction between a compound and a target) and EXPRESSION array
profiles (a measurement of the indirect effects of a compound on the genome). Two
secondary domains, PATHWAY and EXPRESSION STUDY, provide extensions of
these domains. An additional secondary domain, SIGNATURE, provides patterns
of gene expression that can be used to classify compounds, or a set of common
experiments reliably compared to the experiment population, or a defined set of
control experiments.
The DrugMatrix database contains several types of data that require different statistical modeling and handling, including normalization, ratio calculation,
and denominator groupings. While some of the data handling will be discussed
herein, a separate document comprehensively covers this topic (ftp://anonftp.niehs.
nih.gov/ntp-cebs/datatype/Drug_Matrix/DrugMatrix_Calculations.pdf). This document specifically addresses the key computations currently used for different types
of data modeling. A detailed description of raw data handling, such as removal of
no-data spots, exclusion of outlier data points, and data normalization is provided.
Gene expression changes of treated samples compared to controls, together with the
supporting statistics to determine their significance are summarized in detail. The
rationale for using different denominators for ratio calculation of gene expression,
147
Fig. 8.3 A high-level overview of the main information items in the DrugMatrix data warehouse
and user interface. The major information domains used by the DrugMatrix GUI are shown in gold,
green, orange, and blue rectangles. This represents a highly simplified in that it represents only 13
of the 70+ tables in the warehouse
The DrugMatrix relational database is available for download (PostgreSql scripts;
see below) and will be accessible soon through a database client. A Readme will be
available through CEBS which will provide details on how to connect to DrugMatrix
PostgreSql Database.
The two primary data domains, GENE and COMPOUND, are connected via
annotation of chemogenomic-based secondary domains: ASSAY activities (a direct
measure of interaction between a compound and a target) and EXPRESSION array
profiles (a measurement of the indirect effects of a compound on the genome). Two
secondary domains, PATHWAY and EXPRESSION STUDY, provide extensions of
these domains. An additional secondary domain, SIGNATURE, provides patterns
of gene expression that can be used to classify compounds, or a set of common
experiments reliably compared to the experiment population, or a defined set of
control experiments.
The DrugMatrix database contains several types of data that require different statistical modeling and handling, including normalization, ratio calculation,
and denominator groupings. While some of the data handling will be discussed
herein, a separate document comprehensively covers this topic (ftp://anonftp.niehs.
nih.gov/ntp-cebs/datatype/Drug_Matrix/DrugMatrix_Calculations.pdf). This document specifically addresses the key computations currently used for different types
of data modeling. A detailed description of raw data handling, such as removal of
no-data spots, exclusion of outlier data points, and data normalization is provided.
Gene expression changes of treated samples compared to controls, together with the
supporting statistics to determine their significance are summarized in detail. The
rationale for using different denominators for ratio calculation of gene expression,