Challenges in detecting and quantifying intron retention from next generation sequencing data