[Paper Review] Construction of cyclic DNA codes over the Ring $\Z_4[u]/\langle u^2-1 angle $ Based on the deletion distance

Sukhamoy Pattanayak, Abhay Kumar Singh|arXiv (Cornell University)|Mar 13, 2016

DNA and Biological Computing9 references3 citations

TL;DR

This paper constructs cyclic DNA codes of odd length over the ring $\mathbb{Z}_4[u]/\langle u^2-1\rangle$ using deletion distance as a metric, establishing a 1-1 correspondence between ring elements and DNA nucleotide pairs. The key contribution is proving that the deletion distance of a code equals the deletion distance of its subcode generated by $1+u$, enabling efficient design of DNA codes with reverse and reverse-complement constraints, optimal GC-content, and error-correcting capability.

ABSTRACT

In this paper, we develop the theory for constructing DNA cyclic codes of odd length over $R=\Z_4[u]/\langle u^2-1 angle$ based on the deletion distance. Firstly, we relate DNA pairs with a special 16 elements of ring $R$. Cyclic codes of odd length over $R$ satisfy the reverse constraint and the reverse-complement constraint are discussed in this paper. We also study the $GC$-content of these codes and their deletion distance. The paper concludes with some examples of cyclic DNA codes with $GC$-content and their respective deletion distance.

Motivation & Objective

• To develop a theoretical framework for constructing cyclic DNA codes of odd length over the non-chain ring $R = \mathbb{Z}_4[u]/\langle u^2-1\rangle$.
• To ensure DNA codes satisfy reverse and reverse-complement constraints to prevent undesirable hybridization.
• To analyze and control GC-content for thermodynamic stability in DNA-based computing.
• To define and compute deletion distance as a more suitable metric than Hamming distance for DNA sequence errors.
• To demonstrate the construction with explicit examples of codes having specific lengths, GC-content, and deletion-correcting capability.

Proposed method

• Establish a 1-1 correspondence $\theta$ between the 16 elements of the ring $R = \mathbb{Z}_4[u]/\langle u^2-1\rangle$ and DNA nucleotide pairs (A,T,G,C).
• Define cyclic DNA codes over $R$ as ideals in the quotient ring $R[x]/\langle x^n - 1 \rangle$ for odd $n$.
• Impose reverse constraint by requiring generator polynomials to be self-reciprocal and satisfy $x^i g_2^*(x) = g_2(x)$.
• Impose reverse-complement constraint by ensuring $g_3(x)$ divides $ (1+u)x^i g_2^*(x) + (1+u)g_2(x) $, where $g_1(x) = g_2(x)f_1f_2$.
• Use the map $\Phi$ to lift codes from $R^n$ to DNA sequences over $\{A,T,G,C\}^n$, preserving structural and distance properties.
• Prove that the deletion distance $D$ of a code $C$ equals the deletion distance $D_{1+u}$ of its subcode $C_{1+u}$, enabling efficient computation via subcode analysis.

Experimental results

Research questions

• RQ1 How can cyclic DNA codes of odd length be systematically constructed over the non-chain ring $\mathbb{Z}_4[u]/\langle u^2-1\rangle$ while satisfying reverse and reverse-complement constraints?
• RQ2 What is the relationship between the deletion distance of a code and the deletion distance of its subcode generated by $1+u$?
• RQ3 How can GC-content be controlled and computed in cyclic DNA codes over $R$?
• RQ4 What is the impact of the ring structure $\mathbb{Z}_4[u]/\langle u^2-1\rangle$ on the design and error-correcting capability of DNA codes?
• RQ5 Can deletion distance serve as a more effective metric than Hamming distance for modeling DNA sequence errors?

Key findings

• The deletion distance $D$ of a cyclic DNA code $C$ over $R$ is equal to the deletion distance $D_{1+u}$ of its subcode $C_{1+u}$, enabling efficient computation via subcode analysis.
• For a code $C = \langle g_1(x) + (1+u)g_2(x), (1+u)g_3(x) \rangle$, the deletion distance $D$ is determined by the deletion distance of $C_{1+u}$, which is derived from the generator $g_3(x)$.
• Example 6.1 shows a $(6,2)$ DNA cyclic code with 16 codewords, minimum Hamming distance 3, and deletion distance $D=2$, correcting up to 1 deletion error.
• Example 6.2 presents an $(18,8)$ DNA cyclic code with 16 codewords, minimum Hamming distance 9, and deletion distance $D=8$, capable of correcting up to 8 deletion errors.
• The GC-content of the codes is preserved under the map $\Phi$, and codes can be designed to have uniform or controlled GC-content for thermodynamic stability.
• The constructed codes satisfy both reverse and reverse-complement constraints, ensuring no codeword is equal to its reverse or reverse-complement, preventing spurious hybridization.