I've used for Moldavian an average of some samples I find more relevant (about 10 samples both from RM and RO), outliers excluded:
Moldavian,23.542,27.288,16.546,10.864,16.132,1.706 ,0.3,0.848,1.9975,0.418,0.396,0.054,0.032
distance is not ideal (normal for mixed people but can also point to a different source we can't figure out):
Target: Tsurkanka
Distance: 4.3030% / 4.30296412
53.8 Moldavian
46.2 Belorussian
Target: Tsurkanka
Distance: 5.6200% / 5.61997732
51.2 Southwest_Russian
48.8 Moldavian
Target: Tsurkanka
Distance: 6.1322% / 6.13217471
56.4 Ukrainian_Lviv
38.8 Moldavian
4.8 Ukrainian_Belgorod
when I place Belorussian in the source it gets all the % against Ukrainian/Russian/Slovak/Lithuanian. so Belorussian (north Ukrainian?) seems best proxy for about half of her ancestry

