DNA assembly method for a non-model organism using a more distantly-related reference sequence

A reference-guided algorithm for DNA assembly of non-model organisms is proposed. The method enables the use of not only within or closely-related species but also a more distantly-related species as a starting point to help to position reads using the longest common substring algorithm. The performance of the algorithm is evaluated by experiments into two parts. First, shotgun sequences of a known species are simulated as new species to verify the performance of the algorithm, where the references are arbitrarily selected. Second, the effects of distances between the reference and the target sequences are studied. A known sequence with noise injection is represented as a reference for assembling its shotgun sequences. BLAST is used as a quality measurement between the resulting sequence and the ground truth. The experimental results showed that when reference sequences and target sequences are in the same genus, query cover and percent identity are 99.84% and 99.89% respectively; and when they are in the same order, query cover and percent identity are 85.21% and 94.69% respectively. The algorithm can stand the differences between the reference sequences and target sequences up to 20% before the query cover and percent identity reduce to 95%.