Skip to content

Benchmarking - #2

Merged
jeromekelleher merged 3 commits into
tskit-dev:mainfrom
jeromekelleher:benchmarking
Oct 16, 2026
Merged

jeromekelleher merged 3 commits into
tskit-dev:mainfrom
jeromekelleher:benchmarking

Conversation

@jeromekelleher

Copy link
Copy Markdown
Member

No description provided.

Allow recurrent and multiallelic mutations in MatcherIndexes. The C
library takes optional per-site ancestral states and per-mutation
derived states, and validates them against num_alleles; without them
the behaviour is unchanged (ancestral 0, derived 1). The Python
MatcherIndexes wrapper encodes states from the tree sequence via the
new alleles argument.
Scripts to prepare the sc2ts ARG for both engines, generate copy and
synthetic recombinant queries with known ground truth, and run the
tsinfer and tskit Viterbi engines over them, recording timings, paths
and comparable path likelihoods.
Make nested sub-ARGs of the benchmark ARG, each the ARG as of an
earlier date: all nodes at least as old as a time cutoff, with ancestry
no longer above any sample pruned. All sites are kept so a query set
can run against every sub-ARG, and node IDs are reported as full-ARG
IDs so results compare across sizes. The README records the findings
from the first scaling runs, including tskit-hmm's suboptimal results
with internal samples.
@jeromekelleher
jeromekelleher merged commit 15163cb into tskit-dev:main Oct 16, 2026
10 checks passed
@jeromekelleher
jeromekelleher deleted the benchmarking branch October 16, 2026 11:44
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant