Summary and Schedule

This lesson presents basic machine learning concepts featured in Data Analysis for the Life Sciences by Rafael A. Irizarry and Michael I. Love. The lesson is adapted from software for Chapter 9: Practical Machine Learning, which is published under this MIT license. Adaptation was funded by NIH grant 1R25GM141520 awarded to Dr. Gary Churchill at The Jackson Laboratory.

	Setup Instructions	Download files required for the lesson
Duration: 00h 00m	1. Basic Machine Learning	What is machine learning? How is machine learning used in biomedical studies?
Duration: 00h 10m	2. Clustering	How can clusters within high dimensional data can be discovered? ::::::
Duration: 01h 10m	3. Conditional Probabilities and Expectations	How do we get statistical information from specific subsets in our data? ::::::
Duration: 02h 10m	4. Smoothing	Can a model be fitted to a dataset which shape is unknown but smooth? ::::::
Duration: 03h 10m	5. Class Prediction	What is machine learning (ML)? Why should we learn ML? ::::::
Duration: 04h 10m	6. Cross-validation	How can the best configuration of parameters be selected for a machine learning model using only the data available? ::::::
Duration: 05h 10m	Finish

The actual schedule may vary slightly depending on the topics and exercises chosen by the instructor.

Project organization

---

1. Start RStudio.

2. Create a new project in your Desktop called ml-biomed.

• Click the File menu button, then New Project.
• Click New Directory.
• Click New Project.
• Type ml-biomed as the directory name. Browse to your Desktop to create the project there.
• Click the Create Project button.

3. Use the Files tab to create a data folder to hold the data, a scripts folder to hold your scripts, and a results folder to hold results. Alternatively, you can use the R console to run the following commands for step 3 only. You still need to create a project with step 2.

   dir.create("./data")
   dir.create("./scripts")
   dir.create("./results")

Data Sets

---

Download the tissue gene expression data directly from Github and place them in your new data directory.

The data represent RNA expression levels for eight tissues, each with several individuals.

Software Setup

---

1. Install R packages and load the libraries.

   install.packages("rafalib", "RColorBrewer", "gplots",
                   "UsingR", "class", "caret")
   library(rafalib)
   library(RColorBrewer)
   library(gplots)
   library(UsingR)
   library(class)
   library(caret)

2. Install and load packages from Bioconductor.

   if (!require("BiocManager", quietly = TRUE))
        install.packages("BiocManager")
   
   BiocManager::install(c("genefilter", "Biobase", "SpikeIn", "hgu95acdf"))
   library(genefilter)
   library(Biobase)
   library(SpikeIn)
   library(hgu95acdf)
   data(SpikeIn95)