TY - JOUR
T1 - AnnotCompute
T2 - Annotation-based exploration and meta-analysis of genomics experiments
AU - Zheng, Jie
AU - Stoyanovich, Julia
AU - Manduchi, Elisabetta
AU - Liu, Junmin
AU - Stoeckert, Christian J.
PY - 2011
Y1 - 2011
N2 - The ever-increasing scale of biological data sets, particularly those arising in the context of high-throughput technologies, requires the development of rich data exploration tools. In this article, we present AnnotCompute, an information discovery platform for repositories of functional genomics experiments such as ArrayExpress. Our system leverages semantic annotations of functional genomics experiments with controlled vocabulary and ontology terms, such as those from the MGED Ontology, to compute conceptual dissimilarities between pairs of experiments. These dissimilarities are then used to support two types of exploratory analysis-clustering and query-by-example. We show that our proposed dissimilarity measures correspond to a user's intuition about conceptual dissimilarity, and can be used to support effective query-by-example. We also evaluate the quality of clustering based on these measures. While AnnotCompute can support a richer data exploration experience, its effectiveness is limited in some cases, due to the quality of available annotations. Nonetheless, tools such as AnnotCompute may provide an incentive for richer annotations of experiments. Code is available for download at http://www.cbil.upenn.edu/downloads/AnnotCompute.
AB - The ever-increasing scale of biological data sets, particularly those arising in the context of high-throughput technologies, requires the development of rich data exploration tools. In this article, we present AnnotCompute, an information discovery platform for repositories of functional genomics experiments such as ArrayExpress. Our system leverages semantic annotations of functional genomics experiments with controlled vocabulary and ontology terms, such as those from the MGED Ontology, to compute conceptual dissimilarities between pairs of experiments. These dissimilarities are then used to support two types of exploratory analysis-clustering and query-by-example. We show that our proposed dissimilarity measures correspond to a user's intuition about conceptual dissimilarity, and can be used to support effective query-by-example. We also evaluate the quality of clustering based on these measures. While AnnotCompute can support a richer data exploration experience, its effectiveness is limited in some cases, due to the quality of available annotations. Nonetheless, tools such as AnnotCompute may provide an incentive for richer annotations of experiments. Code is available for download at http://www.cbil.upenn.edu/downloads/AnnotCompute.
UR - http://www.scopus.com/inward/record.url?scp=84859732557&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84859732557&partnerID=8YFLogxK
U2 - 10.1093/database/bar045
DO - 10.1093/database/bar045
M3 - Article
C2 - 22190598
AN - SCOPUS:84859732557
SN - 1758-0463
VL - 2011
JO - Database
JF - Database
M1 - bar045
ER -