Abstract
Stochastic, meta-heuristic and linear construction algorithms for the design of DNA strands satisfying Hamming distance and reverse-complement constraints often use a GC-content constraint to pre-process the DNA strands. Since GC-content is a poor predictor of DNA strand hybridization strength the strands can be filtered by post-processing using thermodynamic calculations. An alternative approach is considered here, where the algorithms are modified to remove consideration of GC-content and rely on postprocessing alone to obtain large sets of DNA strands with satisfactory melting temperatures. The two approaches (pre-processing GC-content and post-processing melting temperatures) are compared and are shown to be complementary when large DNA sets are desired. In particular, the second approach can give significant improvements when linear constructions are used.
Original language | English |
---|---|
Pages (from-to) | 441 - 452 |
Number of pages | 11 |
Journal | IEEE/ACM Transactions on Computational Biology and Bioinformatics |
Volume | 11 |
Issue number | 2 |
DOIs | |
Publication status | Published - 21 May 2014 |
Keywords
- dna design
- linear codes
- stochastic local search
- hamming distance
- reverse-complement