search

hammerlab / variant-calling-benchmarks

Automated and curated variant calling benchmarks for Guacamole
Apache License 2.0
2 stars 1 forks source link

no good calls in local-wgs1 benchmark #13

Closed timodonnell closed 8 years ago

timodonnell commented 8 years ago

For the local-wgs1 benchmark, we seem to have no variants in common between the published variants and the guacamole called variants. This is a bug somewhere.

(analysis-venv-2.7)[tim@Tims-MacBook ~/sinai/git/variant-calling-benchmarks]$ cat results/summary.local-wgs1.csv
stat,comparison_dataset,numerator,denominator,percent
calls,,10,,
calls before filtering,,392,,
calls,published,322,,
sensitivity with filters,published,0,322.0,0.0
specificity with filters,published,0,10.0,0.0
sensitivity from pooled calling only without filters,published,0,322.0,0.0
sensitivity individual calling only without filters,published,0,322.0,0.0
specificity without filters,published,0,392.0,0.0
specificity both pooled and individual triggers firing,published,0,373.0,0.0
arahuja commented 8 years ago

In guacamole or in the comparison code?

timodonnell commented 8 years ago

Most likely in the comparison code On Jul 13, 2016 9:44 AM, "Arun Ahuja" notifications@github.com wrote:

In guacamole or in the comparison code?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/hammerlab/variant-calling-benchmarks/issues/13#issuecomment-232359041, or mute the thread https://github.com/notifications/unsubscribe/AAcjuFPvO6l-bymZy6t7iLvr3tjYbmotks5qVOu8gaJpZM4JLbJC .