Help! My reads aren't aligning!

[kclem]

In your use of CRISPResso, you may come across some samples with low alignment rates. This guide will cover what could be causing the low alignment rates, and some troubleshooting steps to help you figure out what is going wrong.

A high alignment rate is critical for accurate CRISPR editing quantification. For this reason, the first figure in the report shows the alignment rate:

The left bar shows the number of reads in the input fastq.

• If this bar is low, it means that too few reads were produced by the sequencer. We usually recommend at least 10,000 reads for a high-quality analysis of genome editing.

The center bar shows the number of reads after preprocessing (adapter trimming (using Trimmomatic), read merging (using FLASH), and quality filtering). CRISPResso will attempt to align these reads to the amplicon sequences.

• Troubleshooting a drop in the number of reads between the first and second bar:
  • Trimmomatic discarded reads. The trimming settings are set using the --trimmomatic_options_string setting in CRISPResso. In particular, the MINLEN and AVGQUAL Trimmomatic settings are used, reads that are too short or have low quality are discarded.
    • Check the Trimmomatic output in the log file CRISPResso_RUNNING_LOG.txt.
    • Try running Trimmomatic on your input outside of CRISPResso using the command in the CRISPResso_RUNNING_LOG.txt.
  • FLASH was unable to merge paired-end reads.
    • Make sure your R1 and R2 are the correct read pair files.
    • Make sure that your amplicon sequence is covered by your sequencing reads. For example, if your amplicon is 300bp long but you've sequenced with 100bp sequencing reads, FLASH will be unable to merge the reads and you'll have a reduction in post-processed reads. If you suspect this could be the case, run CRISPResso with only your R1 or only your R2 file in single-end mode, with 1/2 of your amplicon given as the reference sequence (or --auto mode with no amplicon sequence).
    • You can adjust FLASH merging using the CRISPResso parameters --min_paired_end_reads_overlap (default: 10 for a 10bp minimum overlap between R1 and R2), --max_paired_end_reads_overlap (default: 100 for a maximum 100bp overlap between R1 and R2) and --stringent_flash_merging (default: false for a lenient flash merging threshold).
    • Check the Flash output in the log file CRISPResso_RUNNING_LOG.txt.
    • Try running Flash on your input outside of CRISPResso using the command in the CRISPResso_RUNNING_LOG.txt.
  • Low-quality reads were filtered by CRISPResso
    • Check that sequencing qualities are high by examining the values in the fastq file, or using a program such as FastQC.
    • Try adjusting the fastq quality filters using the parameters --min_average_read_quality, --min_single_bp_quality. By default, no quality filters are applied.

The right bar shows the number of reads that were successfully aligned to an amplicon in CRISPResso. By default, reads must have 60% homology to a reference sequence for successful alignment, meaning that 60% of the bases must match. So if your amplicon sequence is 100bp long, reads containing deletions more than 40bp long will be discarded.

• Troubleshooting a drop in the number of reads between the second and third bar:
  • Make sure your amplicon sequence is correct. Look at the first few lines of your input fastq sequences and perform manual alignment to your reference using a word editor or an online tool (e.g. https://www.ebi.ac.uk/Tools/psa/emboss_needle/), or run CRISPResso with the --auto parameter to detect possible reference amplicon sequences.
  • If you are using interleaved files (where R1 and R2 reads are in the same file, you must use CRISPResso parameter --split_interleaved_input.
  • You can adjust the minimum alignment threshold using the parameters --default_min_aln_score (default 60 for a minimum of 60% homology between the read and the reference sequence) and --amplicon_min_alignment_score (default: 60 for a minimum of 60% homology between the read and the reference sequence. This parameter is a comma-separated string with a value for each amplicon (order given by comma-separated amplicon sequences)). The --default_min_aln_score setting is applied to all amplicons unless they are specifically set to another value using the --amplicon_min_alignment_score parameter.
  • Try setting --default_min_aln_score 0 and --fastq_output. This will report all alignments (even those with 0 homology) and produce the output file CRISPResso_output.fastq.gz with information for each read in the input fastq including the alignment scores to each amplicon. Use this file to determine whether reads discarded at the 60% threshold should be accepted, and what an acceptable --default_min_aln_score would be.