Partitioning of functional gene expression data using principal points

Bayesian Functional Mixed-effects Models with Grouped Smoothness for Analyzing Time-course Gene Expression Data

Objective:

As a result of the development of microarray technologies, gene expression levels of thousands of genes involved in a given biological process can be measured simultaneously, and it is important to study their temporal behavior to understand their mechanisms. Since the dependence between gene expression levels over time for a given gene is often too complicated to model parametrically, sparse functional data analysis has received an increasing amount of attention for analyzing such data.

Methods:

We propose a new functional mixed-effects model for analyzing time-course gene expression data. Specifically, the model groups individual functions with heterogeneous smoothness. The proposed method utilizes the mixed-effects model representation of penalized splines for both the mean function and the individual functions. Given noninformative or weakly informative priors, Bayesian inference on the proposed models was developed, and Bayesian computation was implemented by using Markov chain Monte Carlo methods.

Results:

The performance of our new model was studied by two simulation studies and illustrated using a yeast cell cycle gene expression dataset. Simulation results suggest that our proposed methods can outperform the previously used methods in terms of the mean integrated squared error. The yeast gene expression data application suggests that the proposed model with two latent groups should be used on this dataset.

Conclusion:

The new Bayesian functional mixed-effects model that assumes multiple groups of functions with different smoothing parameters provides an enhanced approach to analyzing timecourse gene expression data.

No wisdom in the crowd: genome annotation in the era of big data - current status and future prospects

Science and engineering rely on the accumulation and dissemination of knowledge to make discoveries and create new designs. Discovery-driven genome research rests on knowledge passed on via gene annotations. In response to the deluge of sequencing big data, standard annotation practice employs automated procedures that rely on majority rules. We argue this hinders progress through the generation and propagation of errors, leading investigators into blind alleys. More subtly, this inductive process discourages the discovery of novelty, which remains essential in biological research and reflects the nature of biology itself. Annotation systems, rather than being repositories of facts, should be tools that support multiple modes of inference. By combining deduction, induction and abduction, investigators can generate hypotheses when accurate knowledge is extracted from model databases. A key stance is to depart from 'the sequence tells the structure tells the function' fallacy, placing function first. We illustrate our approach with examples of critical or unexpected pathways, using MicroScope to demonstrate how tools can be implemented following the principles we advocate. We end with a challenge to the reader.