In this project, I elaborate an analysis pipeline for RNA-seq data through the creation of appropriate functions called from a single main script .
The aim is therefore not to use functions from the most famous NGS data analysis packages but to create custom ones in order to work at a low level on the primitives.
Each step of the pipeline will be represented by a different function.
- Data normalization
- Selection of differentially expressed genes
- Biological interpretation with functional annotation
- Clustering Expression Profiles (k-means)
The data reproduce an experimental design in which the dynamics of beta cells gene expression is monitored under two conditions:
- activation of insulin secretion ("INS" samples)
- inhibition of insulin secretion ("CTRL" samples)
The data are stored in the file matcount_1095199.RData, matrix 10000x78:
N = 10000 genes monitored
M = 13 time instants
3 biological replicates for each temporal instant
where:
rownames(matcount) = EntrezID of the monitored genes
(http:https://www.ncbi.nlm.nih.gov/gene)
colnames(matcount) = CONDITION.TIME.REPLICATES
CONDITION = {"CTRL","INS"}
TIME = {"T0","T1",...,"T12"}
REPLICATES = {"REPL1","REPL2","REPL3"}
It is possible to use this viewer for a better visualization of the notebook: https://nbviewer.jupyter.org/github/lorrandal/RNA_seq/blob/master/RNA_seq.ipynb