Q-consensus: Read Quality Scores for Consensus Sequences

How do you interpret read Q-scores when you have several reads for the same sequence? Q-consensus is introduced to tackle that question. Q-consensus derives from the concept that more information is gained about a template sequence, the more reads that are provided; both when reads agree and disagree on a called base.

For more information, see draft_v0.3.pdf for a motivation for and derivation of the model.

Note: Q-consensus is a work in progress. See "To-do" section below.

Contents

• draft_v0.3.pdf: The derivation of the Q-consensus model [Draft]
• q-consensus.py: A Python implementation of the Q-consensus model
• read-depth.ipynb: Testing how the model predicts Q-scores to increase when multiple reads agree on a consensus.

To-do

• Handle gaps in alignment (and final refinement of model)
• Optimization of speed
• Set up as python module
• Third party validation