EvolDeeds: joint reconstruction objects

alignment + tree + ancestral history = one state. no casually swapping bits between samples and creating a franken-tree; scoring inputs need to travel together

meeting topic: June 30 covered EvolDeeds, indel models and pairHMMs. The later brainstorming document supplies the more detailed working interface: unaligned protein sequences plus an evolutionary model/parameters enter a sampler; outputs include alignments, phylogenetic trees and reconstructed internal-node sequences.

gp120.fa STOCKHOLM output: Newick tree and aligned sequences.

data model: store each alignment/tree/history sample together. Mixing an alignment from one state with a tree or ancestral reconstruction from another can produce a score that no longer describes a sampled state.

code lead from the brainstorming: EvolMoves repository. A separate scoring example calls node js/scripts/calcscore.js data/lg08evol.json data/tree.nh data/tiny.fa; retain that as an example in the named source, not a verified command for every later repository revision.

review questions: are evolutionary parameters fixed or sampled? What is the probability/score convention? How should uncertainty across multiple plausible histories be shown? The later brainstorm explicitly leaves aspects of the user/model-parameter workflow undecided.