Future Research

Must Do For Deployment to Chemistry Team

• Align molecule and consensus sequences against WT and all variants, retain best PID
• Adjust output of off-target barcode table to fit Joe's input

Possible request for AMP

• Skip single molecule seq and consensus - i.e. combine all feature bc events for a gene and do alignment with all features combined

Future Improvements

• Ability to handle RNA
• Metrics for minimum difference between counts/diversity required to make a base call
• QC Metrics for each base
• Probability scores for resolution of multi-mapped reads and tricky indels
• Implement above to improve resolution of real indel vs inadequate coverage
  • Will improve VCF output
• Alignment parameter tuning
  • Customize settings per probe set?
  • Based on empirical testing
• Indel realignment strategy
  • Flag indels
  • Determine probability of proper indel placement
  • Locally align around area

Variant consensus

• If the consensus is the best way to return a sequence, than an error rate for a target seq will not be reported on a single molecule level.
  • Need to separate out variant from reference consensus for error collapsing regardless of the error calc method
  • Right now I believe error rate to variant bases are ignored, limited to only supervised and not translatable to required outputs.