Data Structure for Efficient Storage of Genome Sequencing Data

Zbog golemih duljina genoma, njihova bilo kakva bioinformatička obrada vrlo je memorijski zahtjevna. Općenito, efikasna pohrana ocjena kvaliteta (engl. Phred quality scores) u radnu memoriju predstavlja veći izazov od pohrane samih sljedova pa se prirodno nameće pitanje njihove kompresije. U ovom radu predstavljamo algoritam kompresije ocjena kvaliteta genetskih sljedova koji smanjuje memoriju potrebnu za njihovu pohranu gotovo pa četiri puta. To postiže sažimanjem svake ocjene kvalitete s dva bita umjesto osam te računanjem povoljnog preslikavanja dvobitnih kôdova natrag u ocjene kvalitete. Također u radu ispitujemo količinu informacije koju kompresija gubi nad stvarnim i simuliranim podatcima te utjecaj njene primjene na točnost alata Racon.Due to the tremendous lengths of nearly all genomes, any computational attempt at their analysis is bound to be very memory intensive. Generally, Phred quality scores prove to be more difficult to store efficiently than genetic sequences, making them prime candidates for compression. This paper presents one such compression algorithm that reduces storage needed for Phred quality scores almost four times. This is achieved by using two bits, for representing a single quality score, instead of eight and by calculating satisfactory mappings of two-bit codes to quality scores. Furthermore, we determine the amount of information lost by our algorithm and how this affects the accuracy of the tool Racon