Haplotype inference algorithm

[yk-tanigawa]

• Need to formulate inference algorithm

[yk-tanigawa]

Reference

• EAGLE2
  • http://www.nature.com/ng/journal/vaop/ncurrent/full/ng.3679.html
  • https://github.com/poruloh/Eagle

[yk-tanigawa]

• EAGLE2 is for phasing, i.e. haplotype inference
  • we already know haplotype (with error)
• we need to work on modeling

[yk-tanigawa]

P(H|D) = P(D|H)P(H)

[yk-tanigawa]

Files

• Genome Inference File
  • quick access
  • visualize
  • downstream analysis
• Error Inference File
  • compressed format

[yk-tanigawa]

We are testing our idea with simplest model:

• https://github.com/rivas-lab/presentations/blob/bf5eac8c08e0537d2a889334b2e67483ed557047/yosuke/20170213_yosuke_notes.pdf

[yk-tanigawa]

For the likelihood, we need to estimate error rate and we have three different models:

• fixed (global) error rate
• read specific error rate
• base specific error rate

Would like to compare the performances of these models.