How to Convert a DNA Sequence to a Time Series

[seanlaw]

If I have a DNA sequence, how might I convert it into a time series so that it can be analyzed with STUMPY?

[seanlaw]

According to page 32 of this tutorial by the original authors of the matrix profile, it is claimed that "it is possible to convert DNA
strings to real-valued time series, in a lossless fashion" by doing something like:

import numpy as np

last = 0
dna_ts = np.full(len(dna_seq), -1.0)
for i, nucleotide in enumerate(range(len(dna_seq))):
    if nucleotide == "A":
        dna_ts[i] = last + 2
    elif nucleotide == "G":
        dna_ts[i] = last + 1
    elif nucleotide == "C":
        dna_ts[i] = last - 1
    elif nucleotide == "T":
        dna_ts[i] = last - 2
    else:
        dna_ts[i] = last
        print(f'Warning: Unrecognized nucleotide "{nucleotide}" in index location "{i}"')

    last = dna_ts[i]

Then you should be able to compute the matrix profile for this dna_ts using STUMPY:

import stumpy

m = 60_000
mp = stumpy.stump(dna_ts, m)

However, please be warned that there is no meaningful interpretation of what "z-normalized Euclidean distance" means in nucleotide space!