Xiang-Jun Lu's Software

I have written many purposed-oriented programs since 1986 using FORTARN, C, MATLAB and Perl. This document, however, contains only a brief introduction to the two software packages relevant to the analysis and modeling of nucleic acid structures. They were written as part of my Ph.D. project and should be of general interest, especially SCHNAP.

SCHNAP DNA_Builder References

SCHNAP (***** Now superseded by 3DNA *****)

Introduction

SCHNAP is a software package for analyzing (SCHNAaP) and rebuilding (SCHNArP) nucleic acid structures. It is based on the CEHS scheme proposed by Calldine and El Hassan for working out the local base-pair and step parameters as recommended by the ``Cambridge Convention''. The software was written first in MATLAB and then changed to ANSI C while I was a Ph.D. student working with Prof. C. A. Hunter at the University of Sheffield, and thus, in some sense, SCHNAP represents a joint product of the then Sheffield-Cambridge groups.

Why bother?

Needless to say, there are a number of existing programs for analyzing DNA structures as reviewed by Olson, most noticeably Curves from Lavery, FreeHelix/NewHelix from Dickerson and the Babcock-Olson package. Curves calculates a large number of parameters based on an optimized and mostly ``curved'' global axis. Although local step parameters are also available, local base pair parameters, such as propeller and buckle, are not calculated. Moreover, some parameters--slide, x-displacement and rise (in some circumstances)--from Curves are significantly different from those calculated by other methods. The CEHS scheme is rigorous and reversible as is the Babcock-Olson method, mathematically simple and geometrically meaningful as is FreeHelix/NewHelix. However, CEHS was originally incorporated into NewHelix, which, from a programmer's point of view, is not well organized although there are some valuable ideas in it.

For my Ph.D. project, on DNA base-stacking interactions, I needed to construct a dinucleotide step given a set of local base-pair and step parameters. For a rigorous comparison between my theoretical calculations and X-ray oligonucleotide data, I studied CEHS and NewHelix thoroughly and checked my results with Dr. Mustafa El Hassan. It then appeared to me that a general-purpose analysis program that incorporates good ideas from both CEHS and NewHelix could prove very useful, thus SCHNAaP. On the other hand, SCHNArP was developed as a handy tool for studying sequence-dependent DNA structures, such as a comparison of various DNA bending models. With the new functionality of building a DNA structure with the backbone, it could be very useful for generating arbitrary initial structures in molecular mechanics calculations and molecular dynamics simulations of nucleic acids.

Overall, SCHNAP is not yet another program for analyzing and rebuilding nucleic acid structures but a computer implementation of the elegant CEHS scheme with extensions, such as a new set of global parameters. Its two parts, SCHNAaP and SCHNArP, were designed to form a complete circle, thus make it easy to verify each other. The ordering of atoms in SCHNArP generated structures follows the PDB convention, which makes the least-square fittings with experimental data straightforward. The analysis part, SCHNAaP, can be regarded as a supersede of FreeHelix/NewHelix.

Examples

Here is a detailed listing of structural parameters calculated by SCHNAaP for the well-known Drew-Dickerson dodecamer (PDB code: 1bna), d(CGCGAATTCGCG)2. Starting from the filename.mst output file, the ``standardized'' base-pair step stacking diagrams can be generated as illustrated below using dinucleotide steps 4-6 as an example:

Base-stacking diagrams

SCHNArP can be used to generate the base pair schematic diagram introduced by Calladine and Drew in their DNA work,

Calladine & Drew DNA

or a full atomic DNA structure, such as the U-turn DNA in the IHF-DNA complex.

IHF-DNA Complex

The mathematical rigor of the SCHNAP/CEHS methodology can be further verified by the super-helical nucleosome DNA structure complexed with core particle, which was recently solved by Richmond and co-workers. The 146bp DNA structure was analyzed with SCHNAaP, and reconstructed with SCHNArP using the local CEHS parameters. The RMS deviation for base atoms between the original X-ray structure and SCHNArP generated one is virtually nil.

146bp DNA superhelix

DNA_Builder

DNA_Builder is a program for constructing full three-dimensional double-helical DNA structures based on dinucleotide building blocks. It consists of a dinucleotide structure database, a search/match algorithm, and an easy method for connecting dinucleotides. Conceptually very simple, the program works well for a number of tested cases, and the idea implemented in it is applicable to larger building blocks as well.

References