hummingbird

hummingbird detects differentially methylated regions using a Bayesian hidden Markov model that incorporates location dependence among genomic loci.

Key Features:

• Location-dependent modeling: The Bayesian hidden Markov model incorporates location dependence among genomic loci, unlike methods that assume independence among observations.
• Chromosome-wide information sharing: Bayesian priors permit information sharing across an entire chromosome for improved power of detection.
• Direct state sequence output: The software outputs the best sequence of methylation states, eliminating the need for a subjective p-value threshold to determine significance.
• No replication requirement: The methodology does not require replication in either or both of the two comparison groups.

Scientific Applications:

• Differential methylation detection: Identifies differentially methylated regions between two comparison groups in genomic data.

Methodology:

The method uses a Bayesian hidden Markov model with Bayesian priors applied across an entire chromosome.