Change to Phyre2 output...

[Powell, Harold]

Hi

This is something that most people may not notice, and if they do, I hope that it will be a welcome change.

Previously, the results from Phyre2 have been ordered according to the E-values calculated in HHSearch64, and this has meant that (sometimes) solutions from templates with high sequence identity have appeared lower in the list than solutions with lower seq id. In practice, it can be confusing if the top hit has a seq id of (say) 30%, and a hit placed at #8 in the list has a seq id of 100%.

The change will mean that the % seq id will be taken into account in the output for the top 20 solutions only - these are the ones for which Phyre2 produces a model. So all the models that would have been produced previously will still be produced, but the top hit might be different and the order of the top 20 might be different.

I plan to make the change sometime later this afternoon...

[Powell, Harold]

What I should add is that in my in-house testing I sometimes get identical results using both old and new methods, but sometimes not.

Overall, automated testing at Cameo3D indicates that the top model using the new method is often a better choice, with better statistics like lDDT etc. I haven't seen any results with worse statistics.

Harry