Similarity scores and sequence length