Position Weight Matrix

This tutorial shows how to use rollable functions. Reading a motif-pwm from file, a PositionWeightMatrix function is created using the appropriate alphabet and counts. Since PositionWeightMatrix is a RollableFunction subclass it has a rolling_window method that applies the pwm to all valid windows in the sequence set.

import bionumpy as bnp
import matplotlib.pyplot as plt
from bionumpy.jaspar import read_jaspar_matrix
from bionumpy.position_weight_matrix import PositionWeightMatrix, pwm_from_counts
from bionumpy.encodings.alphabet_encoding import get_alphabet_array_class

# Read the alphabet and counts from jaspar file
alphabet, matrix = read_jaspar_matrix("example_data/MA0080.1.jaspar")

# Convert counts to position weight matrix
pwm = pwm_from_counts(matrix)

# Make an array-class for the alphabet
arrayclass = get_alphabet_array_class(alphabet)

# Get the motif score function
motif_score = PositionWeightMatrix(pwm, arrayclass)

#Get reads
entries = bnp.open("example_data/big.fq.gz").read()

# Calculate the motif score for each valid window
scores = motif_score.rolling_window(entries.sequence)

# Get a histogram of the max-score for each read
plt.hist(scores.max(axis=-1)); plt.show()