One of the challenges in a de-novo puzzle is how to get the sheets aligned. The strategy described here relies on freezing the sheets, banding the segments that are likely to form hydrogen bonds, and using wiggle to gradually move the sheets into alignment.
The slideshow demonstrates the steps in applying this strategy.
Initially, the sheets don't line up very straight.
Freeze the sheets and start banding them.
Wiggle the protein.
As the protein starts to fold, add more bands.
Then make the helices.
How to get the helix to settle without upsetting the sheets while still getting the sheets to link up? Join sheets before making the helices: when the hydrogen bonds are made they'll become more stable. You can always freeze all the packs of sheets and the connecting loops to move the other side.
Try to have all hydrophobic amino acids (orange) inward-facing and hydrophilic amino acids outward-facing. Align the sheets so that the side with more orange faces the inside of the protein.
As an exception, tryptophan segments are orange, but have mixed hydrophobicity, so they are often found outside, on blue side. Similarly, phenylalanine segments are often inside (orange).
[Edited from #global comments by itskimo and Marie]