A Compression Algorithm for Nucleotide Data Based on Differential Direct Coding and Variable Length Look up Table (LUT)

The ongoing exponential increase of genomic data, together with full diploid human genomes, creates new challenges not only for understanding genomic structure, function and development, but also for the storage, navigation and privacy of genomic data. In this paper, we have proposed a modified Direct Differential Coding algorithm. It is a general purposed nucleotide compression algorithm based on variable length LUT. Here the method identifies repeat regions in the individual sequence and the repeat regions are store in the lookup table (LUT). This algorithm compresses both repeat and non repeat sequences. It also handles the non base character and compresses any nucleotide sequences. It gives better result as compared to existing algorithm. The Differential Direct Coding algorithm was a fixed size lookup table algorithm i.e. it used a table of fixed size containing the 64 maximum possible combinations of the triplets obtained by combination of four characters A, G, T and C. We make this table of variable length by adding some more combinations in the look-up table, which are of the size of multiple of triplet i.e. their size is (6,9.12....) since the number of ACSII characters available were not utilized completely. Our algorithm is based on longest common substitution (LCS). It searches a longest common sequence in multiple of 3 and then substitutes an ASCII value in the place of that sequence to generate variable length LUT. In the previous algorithms, the compression ratio so obtained was smaller as compared to the variable length LUT compression algorithm which creates a relatively massive difference when the algorithm is applied on the large genomic repositories. In addition to this, our algorithm also utilizes the maximum number of ASCII characters which are available, thus increasing the efficiency. BACKGROUND The area dealing with the storage of the biological data of living organisms, forces us to use the database management system to store the data. The basic need is to warehouse this data, which carries the sequences of large sizes, lying in the databases. Genbank is one of the biggest databases for biological sequences, whose size roughly doubles in every 18 months [1]. Though in present scenario, space availability is not considered as a big problem, since the high capacity storage devices are easily available in low costs, still the compression of these biological data is of great concern due to many factors like fast searching and retrieval of the data and for performing operations on them. There are many methods to achieve the compression of the data. For e.g. pointer method, table method, etc. [3, 4, 5, 6, 7]. In this research we will particularly concentrate on the table method. There exist many algorithms based on dictionary method like Ziv-Lampel. There exist some other arithmetic encoding algorithms also like Huffman algorithm. However, these universal text compression algorithms are not suitable for compression of biological sequences as they consider the sequence as a pure text stream. If we talk about the DNA sequences, we know that it deals with only four symbols representing four nucleotide bases {A, C, T, G}. these four symbols could have been modeled as {00, 01, 10, 11} respectively, where we can observe that every nucleotide base occupying 8 bit is made to occupy 2 bits, when encoded in the above mentioned binary form. This could have been one of the most efficient encoding schemes, if and only if there were no other symbols in the sequence, other than A, G, T and C base characters. Here, though the encoding can be done, but main problem will occur during decompression as the binary code of the unexpected symbol like N or S which will definitely match with the binary code of A, G, T and C. An another type of algorithm used for DNA compression is Differential Direct (2D) Coding Algorithm [2], which can overcome this problem by differentiating between the base characters and the unexpected symbols. The 2D coding algorithm uses the group of three characters (triplets), being replaced by some other character. Shortcomings of 2D Algorithm a) As discussed, the algorithm stores all the possible combinations of {A, G, T, C, U}, though some of the combinations are not acceptable and are never used, therefore leaving some of the non-printable ASCII characters unused. Govind Prasad Arya et al, / (IJCSIT) International Journal of Computer Science and Information Technologies, Vol. 3 (3) , 2012, 44114416