Strange results

[sXperfect]

Hi,

So i tried to run the release version of the software (HierarchicalModeller_v1.jar) and i got some strange results.
The data I used is 4DNFI1UEG1HD.hic from rao et. al. experiments stored in 4D Nucleome Database chromosome 22 @25kb.
First, the average Spearman correlation is stuck at 0.0

chr22_log.txt:

Input file: output//4DNFI1UEG1HD_filtered.txt
Convert factor: 0.7
Learning rate: 0.1
AVG RMSE: 22104.39887041615
AVG Spearman correlation IFs vs. Reconstructed Dist: -0.38345217398117665
AVG Spearman correlation WishDist vs. Reconstructed Dist: 0.0

and the conversion factor was chosen based on the Spearman correlation with IFs with smallest value (which in this case should be the most negative one)

ratio.txt

Ratio:0.200000, cor: -0.889004
Ratio:0.300000, cor: -0.847088
Ratio:0.400000, cor: -0.722783
Ratio:0.500000, cor: -0.782274
Ratio:0.600000, cor: -0.700732
Ratio:0.700000, cor: -0.610973
Ratio:0.800000, cor: -0.577862
Ratio:0.900000, cor: -0.521465
Ratio:1.000000, cor: -0.517349
Ratio:1.100000, cor: -0.532804
Ratio:1.200000, cor: -0.522525
Ratio:1.300000, cor: -0.529117
Ratio:1.400000, cor: -0.448351
Ratio:1.500000, cor: -0.461233
Ratio:1.600000, cor: -0.436087
Ratio:1.700000, cor: -0.444622
Ratio:1.800000, cor: -0.403792
Ratio:1.900000, cor: -0.432850
Ratio:2.000000, cor: -0.422923
Ratio:2.100000, cor: -0.421341
Ratio:2.200000, cor: -0.424893
Ratio:2.300000, cor: -0.394216
Ratio:2.400000, cor: -0.415498
Ratio:2.500000, cor: -0.410240
Ratio:2.600000, cor: -0.389275
Ratio:2.700000, cor: -0.423461
Ratio:2.800000, cor: -0.388393
Ratio:2.900000, cor: -0.394459
Ratio:3.000000, cor: -0.373484
Ratio:3.100000, cor: -0.390864
Ratio:3.200000, cor: -0.397645
Ratio:3.300000, cor: -0.412245
Ratio:3.400000, cor: -0.395786
Best ratio: 3.000, correlation: -0.373

Could you help me on this one?

h3dg_data.zip