The propensity values are not in range [-1,1] about dspp-keras

[jandom]

@ktamiola should probably tackle this one.

briefly here from me: we have shifted the original scale from -1,1 by +2, so it's now 1 for beta sheet, 2 for random-coil and 3 for helix. 0 now indicates no data for a given sequence position.

@ErmiaAzarkhalili many thank for reporting this inconsistency, you're obviously correct, we'll get it sorted asap

[ktamiola]

Dear @ErmiaAzarkhalili , indeed I have mistakenly used in our README.md an old functional propensity scale. We have shifted all the data by 2.0 as Keras and Tensorflow can't handle np.nan in loss and gradient computation procedures. Thank you for reporting this serious discrepancy.

@ktamiola @jandom Thanks for the quick update. could you please update the README.md file?

[ktamiola]

Absolutely! The README.md file has been updated already!