Overview

Teaching: 15 min
Exercises: 0 min

Learning Objectives

• Get acquinted with core concepts of RNA-seq

Introduction

Kallisto

Kallisto uses pseudo-alignment for RNA-seq quantification. Previous quantifiers attempted to align reads to the genome to measure transcript expression. The position of each read would have to be checked against each part of the reference before quantification. Tools such as Kallisto, Salmon, and sailfish are part of a new generation using pseudo alignments for greatly improved speeds and accurate results.

Pseudo-alignment is not reference-free. Pseudo-alignments do not depend on aligning reads to each position, but do require a reference to quantify expression levels. Kallisto quantifies reads through a de Bruijn graph. Each read is broken into k-mers, and mapped by points of greatest overlap, or k-mer k-compatibility class, as seen in Fig.1 below. This approach means that quantifications is done at the transcript level instead of gene level as with previous aligners.

An advantage of using kallisto is bootstrapping. Bootstrapping can be used to improve statistical deductions. This is used during differential expression analysis in sleuth. A disadvantage is that as kallisto uses pseudo-alignment, it is not as capable in quantifying unknown or novel transcripts.

Sleuth

Sleuth is used for the differential expression analaysis of transcript quantification. While it can work with the output of other quantifiers, Kallisto and sleuth have been designed to work efficiently together. For example, Sleuth makes use of Kallisto’s bootstraps to leverage technical replicate data in determining differentially expressing genes.

It is important to note that Sleuth is an R package, designed to make use of its graphical capabilities. At the end of this topic, we will introduce you to the graphical options available.

Key Points

• RNA-seq is processed through tools such as Kallisto by alignment-free approaches. This approach has improved speeds over previous approaches requiring alignment