search

NESCent / FossilCalibrations

Fossil calibrations database
http://fossilcalibrations.org
BSD 2-Clause "Simplified" License
14 stars 4 forks source link

Tip-taxa search gives inconsistent (unexpected) results #25

Closed jimallman closed 10 years ago

jimallman commented 10 years ago

Try to reproduce this, then diagnose as needed. The calibration 'POLLY Carnivora' was located with these tips: + Felis + Canis

Here's what happens when we search using the tip-taxa filter(s):

WORKS: Browse 'Felis' finds 'POLLY Carnivora' as nearest
WORKS: Browse 'Canis' finds 'POLLY Carnivora' as nearest
WORKS: Tip Search 'Felis' - 'Canis' 
        yields 'POLLY Carnivora' [matches both]

WORKS: Browse 'lynx' finds 'POLLY Carnivora' as nearest
WORKS: Browse 'Procyon' finds 'POLLY Carnivora' as nearest
FAILS: Tip Search 'lynx' - 'Procyon' 
        misses 'POLLY Carnivora' entirely
FAILS: Tip Search 'Procyon' 
        misses 'POLLY Carnivora' entirely
FAILS: Tip Search 'lynx' - 'Canis' 
        yields 'POLLY Carnivora' [matches Canis ONLY]
FAILS: Tip Search 'Procyon' - 'Canis' 
        yields 'POLLY Carnivora' [matches Canis ONLY]
jimallman commented 10 years ago

OK, closer inspection shows that this is expected behavior for our tip-taxa search.

According to the NCBI taxonomy, Lynx is not a descendant of Felis but a sibling under Felinae. As such, its line of ancestry will not cross Felis or Canis, so 'POLLY Carnivora' won't be found.

Similarly, Procyon is descended from neither Felis nor Canis, so again, won't work as a tip-taxon search given the location defined for 'POLLY Carnivora'.

To get the expected results, we should either locate this calibrated node directly on the eponymous taxon in NCBI + Carnivora ...or (if we can't define a one-sided location) on all of its immediate children: + Caniformia + Feliformia + unclassified Carnivora