Computer database compresses DNA sequences used in medical research

Molecular sequence databases, such as those at EMBL, GenBank, and Entrez contain millions of DNA sequences filling many thousands of gigabytes of computer storage capacity of sequences. With almost every new scientific publication in genetics and related sciences, a new sequence is added and the rate at which the data is accumulating is on the rise.

These sequences play a vital role in medical research, disease diagnosis, and the design and development of new drugs. However,

DNA sequences are comprised of just four different bases labelled A, C, G, and T. Each base can be represented in computer code by a two character binary digit, two bits in other words, A (00), C (01), G (10), and T (11). At first glance, one might imagine that this is the most efficient way to store DNA sequences.

DNA sequences, however, are not random, they contain repeating sections, palindromes, and other features that could be represented by fewer bits than is required to spell out the complete sequence in binary. A repeat pattern could be abbreviated to say the binary equivalent of "six times G" for instance, which would be a few bits shorter than explicitly writing "GGGGGG" in binary. Similarly, palindromes could be abbreviated in code relative to their complementary pattern in the DNA sequence.

Many computer users are familiar with compression software that can remove "redundant" code from a music file -- to produce an mp3 -- or an image -- to make a jpg. However, these compression methods lose information. Less familiar to many users are lossless compression methods such as FLAC for sound files, TIFF for images, and the "zip" format for documents and other files. Lossless compression exploits the repeats, palindromes and patterns present in the digital data to reduce the overall size of the file in question.

Now, Taysir Soliman of the Faculty of Computer and Information, at Assiut University, and colleagues Tarek Gharib, Alshaimaa Abo-Alian, and M.A. El Sharkawy of the Faculty of Computer and Information Sciences, at Ain Shams University, have developed a Lossless Compression Algorithm that works with digitized DNA sequences to reduce the amount of computer storage needed for each sequence.

LCA achieves a better compression ratio than existing compression algorithms for DNA, such as GenCompress, DNACompress, and DNAPack, the team says. The same approach could also be used for protein sequences.

The compression algorithm may also have direct application in DNA research, the team suggests. They are now investigating ways in which the results of the compression might be used to differentiate between sections of a DNA sequence that code for proteins and those in the sequence that do not, so-called non-coding regions.