--- title: "Hands-On Session 2: Quantification & Differential Expression" canonical: "https://help.biobam.com/space/BTP/3493888014/Hands-On%20Session%202%3A%20Quantification%20%26%20Differential%20Expression" format: markdown --- # Introduction ## Dataset [https://drive.google.com/drive/folders/1TImxh7w5ZR0-gcnIwAgzl7hMza_rqwfj?usp=sharing](https://drive.google.com/drive/folders/1TImxh7w5ZR0-gcnIwAgzl7hMza_rqwfj?usp=sharing) ## Goals and Algorithms/Tools - Quantification of Long Reads - IsoQuant - Differential Expression Analysis - edgeR ![image-20240305-075847.png](media://bf99c1fe-1bcb-4439-bbbe-36d01b41c7ea) # Task 1: Identifying and Quantifying isoforms using IsoQuant OmicsBox now offers another alternative to FLAIR and IsoSeq to define transcript models from long-read data: IsoQuant. IsoQuant also has two benefits over FLAIR: it can be run without reference annotations (“discovery mode”; FLAIR can only do this when supporting short reads are also provided), and it can quantify both at gene- as well as isoform-level (FLAIR only provides quantification at isoform-level). In order to demonstrate this alternative approach, we will run the same data (leaving off from our aligned reads in .bam format) through IsoQuant. In the third wizard page, “Algorithm Options”, we will select “PacBio CCS or FLNC reads” as our preset, making sure to switch off the “Full-Length Transcripts” checkbox, as well as changing our “Model Construction” strategy to “Default PacBio”. Again, feel free to explore and experiment with other options, although the defaults should be sufficient for our purposes. ## Questions - The way FLAIR and IsoQuant quantify the abundance of isoforms is by simply counting the number of reads assigned to each isoform. Can you think of any reasons why this approach may be problematic? Are there ways to overcome them? Discuss. # Task 2: Differential Expression Analysis Now that we have a sense of the abundance of genes and isoforms in our data, the next natural step is differential expression analysis. Note that there are many different kinds of differential expression (or usage) analyses, although in this task we will simply be examining differential gene expression (DGE) and differential isoform expression (DIE). OmicsBox offers three options for differential expression analysis, and which to use depends primarily on your experimental design: - The most-used option by our users, edgeR, requires your experimental setup to have biological replicates for each condition you want to examine. - For experiments where there are no replicates available and you would like to compare two conditions with just one sample each, you can use NOISeq. - For time-series based experiments, where each sample constitutes a timestep and you are interested in tracking the expression changes over time, you can use maSigPro. In our bear use case, we are primarily interested in the expression changes in adipose tissue of male bears between their active and hibernation phases. Since we have elected to work with two samples of each of our condition, we can use edgeR with a simple design setup, and we will choose to set “hibernation” as the contrast condition, while “active” will be our reference. By running edgeR, we can now find up- and down-regulated genes or isoform, depending on which data we feed into edgeR. ## Questions - How reliable are our results here? What factors limit the significance of our results and how could we adapt our experiments and/or analysis pipeline to improve them? Discuss. - How can we gain more insights into the nature of these differences we have discovered?
Hint We will answer this question in Hands-On Session 3: Functional Annotation.