Overcoming Bias and Systematic Errors in Next Generation Sequencing Data