# Rosalind: BA2H — Implement DistanceBetweenPatternAndStrings
# https://rosalind.info/problems/ba2h/
#
# Given: A DNA string Pattern and a collection of DNA strings Dna.
# Return: DistanceBetweenPatternAndStrings(Pattern, Dna).

let pattern = "AAA"
let strings = ["TTACCTTAAC", "GATATCTGTC", "ACGGCGTTCG", "CCCTAAAGAG", "CGTCAGAGGT"]

# The distance to one string is the best any window of it can do; the distance
# to the collection is the sum over strings. Best, not total, because a motif
# only has to occur once per string.
fn distance_to(text, motif) {
    range(0, len(text) - len(motif) + 1)
      |> map(|i| hamming_distance(substr(text, i, len(motif)), motif))
      |> min()
}

let total = strings |> map(|s| distance_to(s, pattern)) |> sum()

println("Result:   " + str(total))
println("Expected: 5")

fn test_ba2h_distance_between_pattern_and_strings() {
    assert total == 5, "BA2H: got " + str(total)
    # A pattern present exactly in every string would score zero.
    assert distance_to("AAACCC", "AAA") == 0, "BA2H: an exact occurrence should cost nothing"
}