Analyzing ambiguous alignments

[kclem]

If you have an unexpected number of ambiguous alignments, you can figure out what is going on using these strategies:

1. Use a script to analyze ambiguous alignments here. You can download and run it (python plotAmbiguous.py) on a completed CRISPResso run (run with the '--write_detailed_allele_table' parameter).

Here are the parameters:

  -f CRISPRESSO2_FOLDER, --CRISPResso2_folder CRISPRESSO2_FOLDER
                        CRISPResso output folder to plot ambiguous reads for
  -o OUTPUT_ROOT, --output_root OUTPUT_ROOT
                        Plot output root (should not include '.pdf' or '.png')
  --min_freq MIN_FREQ, --min_frequency_alleles_around_cut_to_plot MIN_FREQ
                        Minimum frequency of alleles to plot
  --max_rows MAX_ROWS, --max_rows_alleles_around_cut_to_plot MAX_ROWS
                        Maximum number of rows to plot
  --plot_cut_point      If set, a line at the cut point will be plotted.
  --save_png            If set, pngs will also be produced (as well as pdfs).

The script will pull out ambiguous reads and plot allele tables showing the alignment to the relevant amplicons.

2. You can also use these flags to visualize ambiguous reads by adding them to both references or the first reference:

• --expand_ambiguous_alignments
  If more than one reference amplicon is given, reads that align to multiple reference amplicons will count equally toward each amplicon. Default behavior is to exclude ambiguous alignments. (default: False)
• --assign_ambiguous_alignments_to_first_reference
  If more than one reference amplicon is given, ambiguous reads that align with the same score to multiple amplicons will be assigned to the first amplicon. Default behavior is to exclude ambiguous alignments. (default: False)

3. Run your command with the --fastq_output flag. This will create a fastq file with annotations for each read with alignments and alignment scores to each amplicon. Using this file you can see the why each read was assigned 'AMBIGUOUS'.