Be careful about experimental design : avoid putting all the
replicates in the same lane, using the same barcode for the
replicates, putting different number of samples in lanes etc...
Non- uniformity of the per base read distribution (Illumina Random
Hexamer Priming bias visible on the 13 first bases)
Bias hierarchy : biological condition >> concentration > run/flowcell> lane
At equivalent expression level, a long gene will have more reads than a short one.
Non random coverage along the transcript.
Multiple hit for some reads alignments.