Ten years of pathway analysis: current approaches and outstanding challenges.

Marylyn Ritchie

Marylyn Ritchie

Marylyn Ritchie

Stephen Turner

Stephen Turner

Stephen Turner

+ 0 Evaluators

Pathway analysis is often used to frame results from high-throughput experiments in a functional context. Scores of methods have been developed that employ a variety of knowledge bases and statistical methods. This excellent review categorizes existing methods into three different classes based on how the algorithm works, discusses strengths and weaknesses of each class, and provides insight on the outstanding challenges in pathway analysis.

High-throughput experiments like gene-expression microarrays and RNA-sequencing experiments often result in lists of genes and some associated statistic (e.g. fold changes and p-values). However, a thorough analysis doesn't end with a list of genes. It's useful to try to frame such results in a functionally meaningful biological context. 'Pathway analysis' is a catch-all term that includes many algorithms for accomplishing this task. This review attempts to categorize all these methods into three classes: over-representation analysis; functional class scoring; and topology-based methods. The review gives a few examples of methods in each class (e.g. gene ontology over-representation analysis with hypergeometric tests; gene-set enrichment analysis as a functional class scoring approach), and discusses the strengths and limitations of each class. The review concludes with a discussion of the outstanding challenges in the field, such as low resolution knowledge bases, incomplete and inaccurate annotation information, and methodological challenges (e.g. how to simulate and evaluate new and existing methodology). Similar reviews (see refs {1-4}) have been published in the past, but there has been a long-standing need for an up-to-date and comprehensive review such as this one.

John Jeremy Rice

John Jeremy Rice

John Jeremy Rice

Raquel Norel

Raquel Norel

Raquel Norel

+ 0 Evaluators

This well-written paper reviews methods for pathway-based analysis of 'omics' data, that is methods that provide functional context for a selected list of biological entities obtained by high throughput experiments; the strengths and weaknesses of the three generations of methods are presented, as well as perspectives on current challenges and future needs.