Mathematical models for optimized analysis of nucleic acids.

摘要

Nucleic acid sequencing is a routine task in modern biology. Sequencing is either performed de novo, where the target nucleic acid sequence is unknown, or in a diagnostic setting, where there is a reference sequence available for comparison. Diagnostic sequencing in this context is also known as resequencing. Virtually all de novo sequencing is still performed using the Sanger method, using primers, polymerases, chain terminating dideoxynucleotides and electrophoresis. However, there are several newer technologies available for diagnostic applications, including mass spectrometry and DNA hybridization arrays. In this thesis I present the theoretical framework for three related methods for nucleic acid analysis that extend the utility of mass spectrometry and hybridization arrays to de novo sequencing. These methods exploit certain unusual patterns and mathematical relationships that are present in DNA itself. In the case of mass spectrometry the mathematical relationships exist between the molecular masses of nucleotides and polynucleotides, while for hybridization arrays, these are base sequence patterns. The three methods are: (1) Forced Mass Modulation, a general method for controlling the distribution of masses in groups of oligo- or polynucleotides; (2) Partial Sequencing by Fragmentation, a method for de novo detection of known subsequences in a target nucleic acid by mass spectrometry; and (3) Inference Sequencing by Hybridization, an optimized design for de novo sequencing using oligonucleotide probe arrays. These three techniques can be implemented using recently developed enzymatic and chemical methods for the manipulation of DNA.