Different results with online tool (2.2.12) vs command line (v2.3.1) for paired end reads

[kfenggg]

Different Results: Online Tool vs Command Line

Issue

I'm experiencing discrepancies in results between:

1. The online tool
2. Command line execution (both with default parameters and when attempting to match online tool parameters)

Details

• Input: Providing f1, f2, and amplicon sequences
• Command line results show:
  • Very low read alignment rates
  • Reduced number of reads after preprocessing
• Possible cause: I suspect this might be related to the upgrade from FLASH to fastp

Question

Could someone help me determine the correct settings to achieve consistent results between the online and command-line versions? Or should I install a version <2.3.0?

[Colelyman]

Hey Kevin,

Thanks for using CRISPResso and excellent question. The online version isn't using the current CLI version (yet), so if you want to reproduce the web version on the CLI you should install version 2.2.12, which is the version of CRISPResso used on the web.

Your probably right that the issue is likely caused by FLASH/fastp. Would you mind sharing which version causes low alignment rates?

Thanks,
Cole

[kfenggg]

Hey Cole,

Thanks for the reply! Downloaded v2.2.12 and am getting consistent results now.

The versions that cause low alignments rates are: CRISPResso version 2.3.1, fastp v0.23.4

Best,
Kev

[Colelyman]

Thanks Kevin, glad that you were able to get consistent results.

We definitely don't want alignment rates to be lower because of poor merging. Do you mind sharing some more details on what the comparative rates with fastp/FLASH are? If it is easier/possible to share the data that works too (and if you would rather not post it publicly you can email it [email protected]).

Thanks again for using CRISPResso!
Cole

[kfenggg]

Here are the alignment rates,

fastp: 100% reads in input, 4.5% reads after preprocessing, 0.7% reads aligned
FLASH: 100% reads in input, 95.6% reads after preprocessing, 86.4% reads aligned

Best,
Kev

[Colelyman]

Wow, you're right, those are vastly different! Would you be able to send the data (or part of it) so that we can investigate what is going on?

Thanks,
Cole