An effort has been underway to develop a system for de novo sequencing of single DNA molecules with very long reads. The system operates by optically detecting the passage of fluorescently tagged DNA bases through a detection zone. A successful system would be revolutionary with respect to speed, read length, cost and minimized laboratory infrastructure. An important part of system development is modeling of the detection process. In particular, predicting the expected error from a set of sequencing parameters is helpful in system design. This paper describes variations on the Smith-Waterman algorithm for subsequence alignment used in a single-molecule detection model. The alignment algorithm is used to check the modeled output sequence generated from a known input sequence. Variations based on reasonable assumptions led to over an order of magnitude improvement in alignment speed.