Methylation of DNA and post-translational modifications to DNA-bound histone proteins influence chromatin structure and gene expression. Specific chemical marks along the genome are read and written by specific protein complexes, in particular chromatin contexts. These context-dependent read/write operations can be thought of as rules that implement a computational program.

I show here that chromatin modification is readily modeled as a massively parallel, random-access, self-modifying stored procedure computer – and that it is computationally universal. Instead of looking for meaning in the individual letters of the histone code, this result suggests that the chromatin modification along the genome at any point is like a snapshot of memory in an ongoing computation, and to understand it we should strive to analyze the program that operates dynamically on chromatin memory.

The Chromatin Computer (CC) formalism is useful both analytically and synthetically. Analytically, modeling and simulations of chromatin modification using the formalism developed here and the partial knowledge we have of the function of chromatin-modifying complexes can help us develop a deeper understanding of biology. Mathematical study of this model will help us understand evolutionary and biological constraints on the histone modification system.

The CC can also be used as an abstract model of computation with which to expression algorithms. It may suggest new approaches for in silico computation. I construct a CC program to solve the same Hamiltonian path problem tackled in the first report of creating a computer out of DNA. I implement a simulator of the CC, and run the Hamiltonian path program on it.

While there are many engineering hurdles to overcome, it is also tempting to imagine building a real computer out of chromatin and engineered components.

While the CC model is simple, it can accommodate many aspects of chromatin biology. In living cells, chromatin interacts with cellular functions in various ways: enzymes remove and place chemical modifications on both histones and DNA; remodeling complexes add, remove and translocate nucleosomes; transcription factors bind specific DNA sequences; signal transduction results in changes in chromatin state; loops form to bring distant parts of chromosomes together; changes in chromatin state result in changes in gene expression, which may feed back to change chromatin state; DNA replication requires chromatin compaction and transfer of chromatin state to daughter cells; RNA processing systems operating on nascent transcripts interact with chromatin, including RNA-directed changes in chromatin state. Most of these interactions can be modeled as read-write processes in the Chromatin Computer. The histone and DNA read/write functions are integral to the basic CC model. The other chromatin-interacting biological processes can be modeled as an expanded “instruction set” of CC read-write rules without altering the key result of universal computability. Rather, the biological instruction set enables efficient computation, allowing living cells to carry out nontrivial computations on a large memory.

The model presented here lays the groundwork for several mathematical, biological and engineering avenues of investigation. Seeing the chromatin state as a snapshot of an evolving computation may help us understand the biology of chromatin modifiers, many of which are implicated in disease.

I would like to point out a few excellent and relevant papers that have recently come to my attention. Rohlf et al (Epigenomics, in press) review various models of chromatin modification dynamics, and cite two other papers that model chromatin modification as “rewrite” rules. Prohaska et al describe the evolution of chromatin modification writers and readers, and the corresponding shift from a structural to an informational role. They cast chromatin modification in terms of a computing system operating as a set of state transitions, and estimate the size of memory provided by cis-regulatory networks as opposed to chromatin modifications. Benecke describes chromatin modification as a second layer of information on top of the genome, and also estimates memory size.