🎉Case Study: Aggregate Salmon Quant

In the last case study, we went through the whole process of mapping 16 bulk RNA-seq samples with salmon, and we got 16 transcript-level quantification tables for each sample. In this page, we want to aggregate the transcript-level counts into gene-level counts (the reason is described in this paper) for further Differentially Expressed Gene (DEG) analysis. In order to do so, we will use a R package called "tximport".

For coding, you will learn two things in this case study:

1. More examples of manipulating table using pandas;

2. How to integrate R seamlessly include some open-box R packages into python.

Step 1. Prepare transcript2gene map

See Jupyter Notebook

In this notebook, we will use the GENCODE GTF file to extract informations related to next step. The same GTF file is used in salmon index when I prepare the data. It is IMPORTANT to keep using the same reference (e.g., genome FASTA, gene annotation GTF) throughout your single project.

We will extract gene_id for each transcript_id from the GTF, then save them into another table.

Step 2. Transcript Quant To Gene Quant

See Jupyter Notebook

In this notebook, we will use the R package "tximport" to aggregate transcript quant into gene quant for each sample. This step also provides an example how we can combine R and python in the same notebook, taking advantages from both world!

I also created one large table in the end for the whole dataset, which contains all samples' gene-level information. This is the start point of all our future analysis, just a 6Mb CSV table.