Machine Learning in Biotechnology and Life Sciences

Saleh Alkhalifa

• Publisher: Packt Publishing
• Year: 2022
• ISBN:
  9781801815673
• Edition: 1
• Topic: Mathématiques
• Subtopic: Probabilités et statistiques

Explore all the tools and templates needed for data scientists to drive success in their biotechnology careers with this comprehensive guideKey Features• Learn the applications of machine learning in biotechnology and life science sectors• Discover exciting real-world applications of deep learning and natural language processing• Understand the general process of deploying models to cloud platforms such as AWS and GCPBook DescriptionThe booming fields of biotechnology and life sciences have seen drastic changes over the last few years. With competition growing in every corner, companies around the globe are looking to data-driven methods such as machine learning to optimize processes and reduce costs. This book helps lab scientists, engineers, and managers to develop a data scientist's mindset by taking a hands-on approach to learning about the applications of machine learning to increase productivity and efficiency in no time.You'll start with a crash course in Python, SQL, and data science to develop and tune sophisticated models from scratch to automate processes and make predictions in the biotechnology and life sciences domain. As you advance, the book covers a number of advanced techniques in machine learning, deep learning, and natural language processing using real-world data.By the end of this machine learning book, you'll be able to build and deploy your own machine learning models to automate processes and make predictions using AWS and GCP.What you will learn• Get started with Python programming and Structured Query Language (SQL)• Develop a machine learning predictive model from scratch using Python• Fine-tune deep learning models to optimize their performance for various tasks• Find out how to deploy, evaluate, and monitor a model in the cloud• Understand how to apply advanced techniques to real-world data• Discover how to use key deep learning methods such as LSTMs and transformersWho this book is forThis book is for data scientists and scientific professionals looking to transcend to the biotechnology domain. Scientific professionals who are already established within the pharmaceutical and biotechnology sectors will find this book useful. A basic understanding of Python programming and beginner-level background in data science conjunction is needed to get the most out of this book.

Section 1: Getting Started with Data

This section describes the basics of Python, SQL, and translating raw data into meaningful visualizations and representations as the first step of a strong data science project. Novice students generally find themselves overwhelmed by the vast amount of data science content found on the internet or in print. This book remedies this issue by focusing on the most important and valuable must-know elements for getting started in the field.

This section comprises the following chapters:

• Chapter 1, Introducing Machine Learning for Biotechnology
• Chapter 2, Introducing Python and the Command Line
• Chapter 3, Getting Started with SQL and Relational Databases
• Chapter 4, Visualizing Data with Python

Chapter 1: Introducing Machine Learning for Biotechnology

How do I get started? This is a question that I have received far too frequently over my last few years as a data scientist and consultant operating in the technology/biotechnology sectors, and the answer to this question never really seemed to change from person to person. My recommendation was generally along the lines of learning Python and data science through online courses and following a few tutorials to get a sense of how things worked. What I found was that the vast majority of scientists and engineers that I have encountered, who are interested in learning data science, tend to get overwhelmed by the large volume of resources and documentation available on the internet. From Getting Started in Python courses to Comprehensive Machine Learning guides, the vast majority of those who ask the question How do I get started? often find themselves confused and demotivated just a few days into their journey. This is especially true for scientists or researchers in the lab who do not usually interact with code, algorithms, or predictive models. Using the Terminal command line for the first time can be unusual, uncomfortable, and – to a certain extent – terrifying to a new user.

This book exists to address this problem. This is a one-stop shop to give scientists, engineers, and everyone in-between a fast and efficient guide to getting started in the beautiful field of data science. If you are not a coder and do not intend to be, you have the option to read this book from cover to cover without ever using Python or any of the hands-on resources. You will still manage to walk away with a strong foundation and understanding of machine learning and its useful capabilities, and what it can bring to the table within your team. If you are a coder, you have the option to follow along on your personal computer and complete all the tutorials we will cover. All of the code within this book is inclusive, connected, and designed to be fully replicable on your device. In addition, all of the code in this book and its associated tutorials is available online for your convenience. The tutorials we will complete can be thought of as blueprints to a certain extent, in the sense that they can be recycled and applied to your data. So, depending on what your expectations of the phrase getting started are, you will be able to use this book effectively and efficiently, regardless of your intent to code. So, how do we plan on getting started?

Throughout this book, we will introduce concepts and tutorials that cater to problems and use cases that are commonly experienced in the technology and biotechnology sectors. Unlike many of the courses and tutorials available online, this book is well-connected, condensed, and chronological, thus offering you a fast and efficient way to get up to speed on data science. In under 400 pages, we will introduce the main concepts and ideas relating to Python, SQL, machine learning, deep learning, natural language processing, and time-series analysis. We will cover some popular approaches, best practices, and important information every data scientist should know. In addition to all of this, we will not only put on our data scientist hats to train and develop several powerful predictive models, but we will also put on our data engineer hats and deploy our models to the cloud using Amazon Web Services (AWS) and Google Cloud Platform (GCP). Whether you are planning to bring data science to your current team, train and deploy the models yourself, or start interviewing for data scientist positions, this book will equip you with the right tools and resources to start your new journey, starting with this first chapter. In the following sections, we will cover a few interesting topics to get us started:

• Understanding the biotechnology field
• Combining biotechnology and machine learning
• Exploring machine learning software

With that in mind, let's look at some of the fun areas within the field of biotechnology that are ripe for exploration when it comes to machine learning.

Understanding the biotechnology field

Biotechnology, as the name suggests, can be thought of as the area of technological research relating to biology when it comes to living organisms or biological systems. First coined in 1919 by Karoly Ereky, the father of biotechnology, the field traditionally encompassed the applications of living organisms for commercial purposes.

Some of the earliest applications of biotechnology throughout human history include the process of fermenting beer, which dates as far back as 6,000 BC, or preparing bread using yeast in 4,000 BC, or even the development of the earliest viral vaccines in the 1700s.

In each of these examples, scientific or engineering processes utilized biological entities to produce goods. This concept was true then and had remained just as true throughout human history. Throughout the 20th century, major innovative advancements were made that changed the course of mankind for the better. In 1928, Alexander Fleming identified a mold that halted the replication of bacteria, thus leading to penicillin – the first antibiotic. Years later, in 1955, Jonas Salk developed the first polio vaccine using mammalian cells. Finally, in 1975 one of the earliest methods for the development of monoclonal antibodies was developed by George Kohler and Cesar Milstein, thus reshaping the field of medicine forever:

Figure 1.1 – A timeline of a few notable events in the history of biotechnology

Toward the end of the 20th century and the beginning of the 21st century, the field of biotechnology expanded to cover a diverse bevy of sub-fields, including genomics, immunology, pharmaceutical treatments, medical devices, diagnostic instruments, and much more, thus steering its focus away from its agricultural applications and more on human health.

Success in Biotech Health

Over the last 20 years, many life-changing treatments and products have been approved by the FDA. Some of the industry's biggest blockbusters include Enbrel® and Humira®, monoclonal antibodies for treating rheumatoid arthritis; Keytruda®, a humanized antibody for treating melanoma and lung cancer; and, finally, Rituxan®, a monoclonal antibody for treating autoimmune diseases and certain types of cancer. These blockbusters are but a sample of the many significant advances that have happened in the field over the past few decades. These developments contributed to creating an industry that's larger than many countries on Earth while changing the lives of millions of patients for the better.

The following is a representation of a monoclonal antibody:

Figure 1.2 – A 3D depiction of a monoclonal antibody

The biotechnology industry today is flourishing with many new and significant advances for treating illnesses, combatting diseases, and ensuring human health. However, with the space advancing as quickly as it is, the discovery of new and novel items is becoming more difficult. A great scientist once told me that advances in the biopharmaceutical industry were once made possible by pipettes, and then they were made possible by automated instruments. However, in the future, they will be made possible by Artificial Intelligence (AI). This brings us to our next topic: machine learning.

Combining biotechnology and machine learning

In recent years, scientific advancements in the field, boosted by applications of machine learning and various predictive technologies, have led to many major accomplishments, such as the discovery of new and novel treatments, faster and more accurate diagnostic tests, greener manufacturing methods, and much more. There are countless areas where machine learning can be applied within the biotechnology sector; however, they can be narrowed down to three general categories:

• Science and Innovation: All things related to the research and development of products.
• Business and Operations: All things related to processes that bring products to market.
• Patients and Human Health: All things related to patient health and consumers.

These three categories are essentially a product pipeline that begins with scientific innovation, where products are brainstormed, followed by business and operations, where the product is manufactured, packaged, and marketed, and finally the patients and consumers that utilize the products. Throughout this book, we will touch on numerous applications of ...

Table of contents

• Machine Learning in Biotechnology and Life Sciences
• Contributors
• Preface
• Section 1: Getting Started with Data
• Chapter 1: Introducing Machine Learning for Biotechnology
• Chapter 2: Introducing Python and the Command Line
• Chapter 3: Getting Started with SQL and Relational Databases
• Chapter 4: Visualizing Data with Python
• Section 2: Developing and Training Models
• Chapter 5: Understanding Machine Learning
• Chapter 6: Unsupervised Machine Learning
• Chapter 7: Supervised Machine Learning
• Chapter 8: Understanding Deep Learning
• Chapter 9: Natural Language Processing
• Chapter 10: Exploring Time Series Analysis
• Section 3: Deploying Models to Users
• Chapter 11: Deploying Models with Flask Applications
• Chapter 12: Deploying Applications to the Cloud
• Other Books You May Enjoy