PANTHER pathway: an ontology-based pathway database coupled with data analysis tools

Abstract

The availability of whole genome sequences from various model organisms and increasing experimental data and literatures stimulated the evolution of a systems approach for biological research. The development of computational tools and algorithms to study biological pathway networks has made great progress in helping analyze research data. Pathway databases become an integral part of such an approach. This chapter first discusses how biological knowledge is represented, particularly the importance of ontologies or standards in systems biology research. Next, we use PANTHER Pathway as an example to illustrate how ontologies and standards play a role in data modeling, data entry, and data display. Last, we describe the usage of such systems. We also describe the computational tools that utilize PANTHER Pathway information to analyze gene expression experimental data.

Figure 1. Two views of pathway reactions

(A) and (C). Reaction diagrams that display activity flows of two activation reactions that are commonly used in scientific papers. (B) and (D). Diagrams drawn in CellDesigner, which captures the different molecular mechanisms of the activation reactions illustrated in (A) and (C), respectively.

Figure 2. Phylogenetic tree of a PANTHER protein family

This figure shows an example of a PANTHER protein family PTHR1537, a voltage-gated potassium channel family. The phylogenetic tree was constructed computationally from sequence data. (A) The tree is collapsed to show the subfamilies as leafnodes (blue diamond nodes). Subfamily nodes correspond to common ancestors of extant family members, and are annotated by expert biologists with their inferred molecular functions and roles in biological processes and pathways, based on experiments performed on extant proteins. Each subfamily is represented by a hidden Markov model (HMM) to allow classification of newly discovered protein sequences. (B) A subset of the tree (under red arrow in (A)) is expanded to show an individual training sequence that is used to build a family multisequence alignment, as well as family and subfamily HMMs.

Figure 2. Phylogenetic tree of a PANTHER protein family

This figure shows an example of a PANTHER protein family PTHR1537, a voltage-gated potassium channel family. The phylogenetic tree was constructed computationally from sequence data. (A) The tree is collapsed to show the subfamilies as leafnodes (blue diamond nodes). Subfamily nodes correspond to common ancestors of extant family members, and are annotated by expert biologists with their inferred molecular functions and roles in biological processes and pathways, based on experiments performed on extant proteins. Each subfamily is represented by a hidden Markov model (HMM) to allow classification of newly discovered protein sequences. (B) A subset of the tree (under red arrow in (A)) is expanded to show an individual training sequence that is used to build a family multisequence alignment, as well as family and subfamily HMMs.

Figure 3. Associating pathway molecule classes to subfamilies in a phylogenetic tree.

(A) A molecule class of phospholipase C (PLC, in yellow) in the Angiotensin II-stimulated signaling through G protein and beta-arrestin (PANTHER accession P05911). (A) The phylogenetic tree of phospholipase C subfamilies (PANTHER accession PTHR10336:SF10 -- SF12). A web curation interface is built to allow expert biologists to associate protein training sequences of the tree to the molecule class. For each annotation, a confidence code must be selected from the list defined by the Gene Ontology Consortium and a PubMed identifier of the source of the literature references as evidence.

Figure 3. Associating pathway molecule classes to subfamilies in a phylogenetic tree.

(A) A molecule class of phospholipase C (PLC, in yellow) in the Angiotensin II-stimulated signaling through G protein and beta-arrestin (PANTHER accession P05911). (A) The phylogenetic tree of phospholipase C subfamilies (PANTHER accession PTHR10336:SF10 -- SF12). A web curation interface is built to allow expert biologists to associate protein training sequences of the tree to the molecule class. For each annotation, a confidence code must be selected from the list defined by the Gene Ontology Consortium and a PubMed identifier of the source of the literature references as evidence.

Figure 4. Gene expression data from Compared gene lists tool viewed on PANTHER website.

(A) The results from Compare gene lists tool. Two sample lists were uploaded to the tool, List_A and List_B. The NCBI human gene list was used as the reference list. (B) The results in the Angiogenesis pathway show molecule classes only present in List_A (red), List_B (green), or both (yellow).

Figure 4. Gene expression data from Compared gene lists tool viewed on PANTHER website.

(A) The results from Compare gene lists tool. Two sample lists were uploaded to the tool, List_A and List_B. The NCBI human gene list was used as the reference list. (B) The results in the Angiogenesis pathway show molecule classes only present in List_A (red), List_B (green), or both (yellow).

Figure 5. Gene expression data from Analyze a list of genes with gene expression values tool viewed on PANTHER website.

(A) The output of the tool with a list of P-values for each comparison between a functional category distribution and the reference distribution. (B) Comparison of the distributions from T-cell activation signaling pathway (red) and reference (blue) in graph view. (C) A pathway diagram of T-cell activation signaling pathway that is visualized using an interactive pathway JAVA applet that colors the pathway using a “heat map” derived from the input values.