Introduction to NumPy

Objectives

• What is NumPy and why it’s important for bioinformatics

• Performance advantages over Python lists

• Foundation for other scientific libraries

Instructor note

• Teaching : 10 min

• Demo: 10 min

What is numpy?

• NumPy is short for “Numerical Python”

• Core python library for scientific computing

• Useful for processing large quantities of same-type data

• Foundation for:

  • Data manipulation, analysis and visualization libraries (Pandas, Matplotlib, scipy)

  • Machine learning libraries (scikit-learn, TensorFlow, PyTorch)

• NumPy operations are written in compiled C, significantly speeding up mathematical operations

Ref: Array programming with NumPy

Why NumPy is Essential for Bioinformatics

• Bioinformatics involves processing and analyzing vast amounts of biological data, from genomic sequences to protein structures

• NumPy’s efficient N-dimensional arrays allow for fast and memory-efficient processing of these large datasets

  • C-optimized operations: NumPy’s compiled C backend makes processing of big-data feasible without having C programming knowledge

  • Reduced memory footprint help optimize the big-data processing: Accelerating Key Bioinformatics Tasks 100-fold by Improving Memory Access

• Run statistical analysis/operations using Biological Data effectively (low barrier to entry)

• Numpy is essential for Python-based machine learning applications on biological datasets

Note

In essence, NumPy bridges the gap between high-level Python programming and the performance requirements of modern bioinformatics, making it possible to analyze the increasingly large datasets generated by modern biological research techniques.

NumPy Arrays vs Python Lists

• Lists are data structures used to store collections of elements

• NumPy arrays enforce a single data type for all elements

• Benefits of NumPy arrays:

  • Homogeneity removes need for type checking during operations

  • Contiguous memory allocation (faster than Python’s scattered storage)

  • Vectorization allows operations on entire arrays without loops

  • Rich set of mathematical functions and operations

Creating NumPy Arrays

Demo

1D Arrays from lists

import numpy as np

# Create from list
py_list = list(range(1,5))
np_array = np.array(py_list)
print(np_array)  # Output: array([1, 2, 3, 4])

2D Arrays (matrices)

# Create a 2D array
rows, cols = 3, 4
list_of_list = [[j for j in range(cols)] for i in range(rows)]
np_array = np.array(list_of_list)
print(np_array)

Output:

[1 2 3 4]

array([[0, 1, 2, 3],
       [0, 1, 2, 3],
       [0, 1, 2, 3]])

Creating arrays from scratch

Demo

# Range of values
print("np.arange")
print(np.arange(1, 10, 2))  # <start,stop,step>, stop is not included in the array

# Arrays of zeros
print("np.zeros")
print(np.zeros((2, 2)))     # Array of zeros

# Arrays of ones
print("np.ones")
print(np.ones(5))           # Array of ones

# 2-D arrays
print("2-D arrays")
print(np.ones([5,2])) # 2-D array of ones 
print(np.random.random((2, 2)))  # 2-D array Random values between 0 and 1

Output

np.arange
[1 3 5 7 9]
np.zeros
[[0. 0.]
 [0. 0.]]
np.ones
[1. 1. 1. 1. 1.]
2-D arrays
[[1. 1.]
 [1. 1.]
 [1. 1.]
 [1. 1.]
 [1. 1.]]
[[0.76528238 0.95473465]
 [0.81062275 0.14766793]]

Examining numpy array structure and storage

• NumPy arrays come with several attributes that provide important information about their structure and data storage.

Attribute	Description	Example	Purpose
shape	A tuple of integers representing the size of each dimension of the array	(3, 4) (2D array with 3 rows and 4 columns)	Understands the layout and number of elements within the array.
ndim	An integer indicating the dimensionality of the array (number of dimensions)	2 (for a 2D array), 1 (for a vector)	Clarifies how many axes are used to access elements.
size	An integer representing the total number of elements within the array	12 (for a 2D array with shape (3, 4))	Provides a quick way to determine the total number of elements.

Demo

print("2-D array")
np_2d = np.ones([5,2])
print("\tshape", np_2d.shape)
print("\tndim", np_2d.ndim)
print("\tsize", np_2d.size)

Output

2-D array
	shape (5, 2)
	ndim 2
	size 10

More info

Additional notes: Exercise

Exercise 1: Creating Arrays from Scratch:

In this exercise, you’ll practice creating NumPy arrays using different built-in functions.

Tasks:

1. Create a 1D array containing integers from 5 to 50 with a step size of 5 using np.arange().

2. Create an array of 8 evenly spaced values between 0 and 1 (inclusive) using np.linspace().

3. Create an array of 10 random integers between 1 and 100 using np.random.randint().

4. Create an array of shape (3,3) filled with the value 3.14 using np.full().

Expected Output:

# After task 1
array([ 5, 10, 15, 20, 25, 30, 35, 40, 45, 50])

# After task 2
array([0.        , 0.14285714, 0.28571429, 0.42857143, 0.57142857,
       0.71428571, 0.85714286, 1.        ])

# After task 3 (your values will differ due to randomness)
array([42, 67, 89, 14, 53, 12, 95, 78, 37, 51])

# After task 4
array([[3.14, 3.14, 3.14],
       [3.14, 3.14, 3.14],
       [3.14, 3.14, 3.14]])

• Test shape, ndim, size attributes of the Numpy arrays created in above tasks

print("Exercise 1: Creating Arrays from Scratch")

# Task 1: Create array from 5 to 50 with step size of 5
array1 = np.arange(5, 51, 5)
print("Task 1 - Array with integers from 5 to 50, step 5:")
print(array1)
print()

# Task 2: Create array of 8 evenly spaced values between 0 and 1
array2 = np.linspace(0, 1, 8)
print("Task 2 - 8 evenly spaced values between 0 and 1:")
print(array2)
print()

# Task 3: Create array of 10 random integers between 1 and 100
array3 = np.random.randint(1, 101, 10)
print("Task 3 - 10 random integers between 1 and 100:")
print(array3)
print()

# Task 4: Create 3x3 array filled with 3.14
array4 = np.full((3, 3), 3.14)
print("Task 4 - 3x3 array filled with 3.14:")
print(array4)
print("\n" + "-"*50 + "\n")