Parsimonious Higher-Order Hidden Markov Models for Improved Array-CGH Analysis with Applications to Arabidopsis thaliana

Abstract

Array-based comparative genomic hybridization (Array-CGH) is an important technology in molecular biology for the detection of DNA copy number polymorphisms between closely related genomes. Hidden Markov Models (HMMs) are popular tools for the analysis of Array-CGH data, but current methods are only based on first-order HMMs having constrained abilities to model spatial dependencies between measurements of closely adjacent chromosomal regions. Here, we develop parsimonious higher-order HMMs enabling the interpolation between a mixture model ignoring spatial dependencies and a higher-order HMM exhaustively modeling spatial dependencies. We apply parsimonious higher-order HMMs to the analysis of Array-CGH data of the accessions C24 and Col-0 of the model plant Arabidopsis thaliana. We compare these models against first-order HMMs and other existing methods using a reference of known deletions and sequence deviations. We find that parsimonious higher-order HMMs clearly improve the identification of these polymorphisms. Moreover, we perform a functional analysis of identified polymorphisms revealing novel details of genomic differences between C24 and Col-0. Additional model evaluations are done on widely considered Array-CGH data of human cell lines indicating that parsimonious HMMs are also well-suited for the analysis of non-plant specific data. All these results indicate that parsimonious higher-order HMMs are useful for Array-CGH analyses. An implementation of parsimonious higher-order HMMs is available as part of the open source Java library Jstacs (www.jstacs.de/index.php/PHHMM).

Autoregressive Higher-Order Hidden Markov Models: Exploiting Local Chromosomal Dependencies in the Analysis of Tumor Expression Profiles

“…This extends previous findings where standard mixture models have been outperformed by first-order and higher-order HMMs [11], [13], [24]. Moreover, these performances can be further improved by autoregressive higher-order HMMs (AR()-HMM() with and ; non-black curves in Figure 3a).…”

Nuclear retention of the transcription factor NLP7 orchestrates the early response to nitrate in plants

“…Data were averaged on the dye-swap to remove tile-specific dye biases and across biological replicates. Normalized data were then analysed using a three-states Parsimonious Higher-Order Hidden Markov Model (PHHMM, www.jstacs.de/index.php/PHHMM) to model the joint distribution of the þ N and À N-derived signals, taking into account spatial dependency between the DNA sequence reported by the probes on the tiling array 33 . Briefly, normalized log2 þ N/ À N signal ratio for each probe corresponds to the observed values of the Hidden Markov chain, while the hidden chain is the enrichment state of the DNA sequence reported by each probe in the ChIP-chip experiments, respectively.…”

“…First, initial model parameters are iteratively maximized by a specific Bayesian Baum-Welch training algorithm enabling the integration of previous knowledge and by a data-dependent reduction of transition parameters. Second, based on this training, optimized model parameters are obtained that allow the parsimonious higher-order HMM with univariate Gaussian emission distributions to effectively model spatial dependencies between measurements 33 . This HMM approach provides a probabilistic classification of each probe into one of the three states, namely non-enriched, enriched and depleted in the þ N/ À N comparison.…”

Copy Number Variation in Transcriptionally Active Regions of Sexual and Apomictic Boechera Demonstrates Independently Derived Apomictic Lineages

“…All normalized CGH profiles were analyzed by a mixture model of three Gaussian distributions described in Seifert et al (2012) to identify depletions, enrichment, and unchanged regions between a test genotype and the reference sample by specifically adapting the parameters of the mixture model to both CGH profiles (replicates from each dye swap; see Supplemental Figure 1 online) of each test genotype using a Bayesian Expectation Maximization algorithm. Identification of depleted, unchanged, and enriched regions was done by performing a decoding of the measured log2 ratios into the most likely underlying state of each probe (depleted, unchanged, or enriched).…”