Issue solved. The problem wasn't with pipeline but with the fastq files, specifically read 2. The read was truncated in all files. Running another set of samples showed the pipeline worked successfully.