Run Probabilities of Seed-Like Patterns and Identifying Good Transition Seeds

Transition seeds exhibit a good tradeoff between sensitivity and specificity for homology search in both coding and non-coding regions. However, identifying good transition seeds is intractable. This work studies the hit probability of high-order seed patterns. Based on our theoretical results, we propose an efficient method for ranking transition seeds for seed design and list good seeds in different Bernoulli sequence models.