DNA Library Design for Molecular Computation R. Penchovsky and J. Ackermann Journal of Computational Biology v.10, n.2, 2003 Summarized by In-Hee Lee
1. Introduction (1/2) Accuracy of the computation depends on the ability to discriminate between matching hybrids and those with mismatches. Sequence design methods so far based on Hamming and related distances. Ignoring physical property of DNA
1. Introduction (2/2) DNA library design Based on thermodynamic stability of DNA hybrids. To deal with prob. dist. of alternative DNA/DNA duplex, a partition function is computed. Used the Vienna RNA folding package. Also used NN parameters. Random search algorithm.
2. Materials and Methods (1/2) Library constraints 12-bit library, different sequence for different positions and bits. No G’s. Reduce the stability of possible secondary structures and hybridization among the library. Length – 16-mer. No occurrences of four or more consecutive identical nucleotides. No runs of three consecutive C’s at either the 5’ or 3’ end. Maximize the free energy gap btw. the weakest specific hybridization and strongest non-specific hyb. Restricted range of Tm (1.5ºC).
2. Materials and Methods (2/2) DNA library design algorithm Consists of two steps. Step A. Random search algorithm Step B. Construct the 12-bit DNA library. Tested four library sequences 111111111111, 000000000000, 010101010101, 101010101010
Word Sequences
Library Generation Mix-and-split
3. Experimental Results (1/4) The Goodness of generated sequences Set of words Words and the 4 library sequences
3. Experimental Results (2/4) Integrity and accuracy of the library 4 library sequences with ‘zero’ capture probes
3. Experimental Results (3/4) Integrity and accuracy of the library 4 library sequences with ‘one’ capture probes
3. Experimental Results (4/4) Localizing the unexpected amplified product. Separated PCR for 4 library sequences with primer 2: 111111111111 3: 101010101010 4: 000000000000 5: 010101010101
4. Discussion Library design based on thermodynamics. Increasing the set size is not necessarily connected with a reduction of the word quality. More accurate than using the Hamming distances.