The nucleotides are intended to be printed out as perimeters only, so turn off the fill plugin. Use Multiply to print more pieces at once: on a TOM I can print eight at a time (2x2 of the STL pair). I also recommend upgrading to SF44, which does filament retraction better than the default Skeinforge 35. Look up directions for upgrading online, as there are a couple of minor tricks involved.

The pieces have their letter embossed on their outer edges, but I still prefer to print them in different colors to make the sequence obvious. Remember: A's link across to T's and G's link across to C's. The snap shapes enforce this pairing. You may also notice that the A-T link is less secure than the G-C link: this is intentional, to model the fact that G-C pairs link by three bonds while A-T pairs link by only two.

There are some odd angles involved in creating a helix out of flat pieces, so the snap links along the helix backbone are only meant to line up along the outer edge. The letters are oriented such that if you read an assembled sequence from left to right, you'll be reading from the 5' end to the 3' end of the strand, which is the convention for reporting DNA sequences.

The OpenSCAD file is included in case you want to see how I modeled this or make changes. It also has an assembly function that allows you to specify a sequence and it will display the corresponding single helix. You'll have to make a complementary sequence as well to see the full double-helix.

These pieces print out so quickly that it doesn't take long to make a good strand (I can do about 30 base pairs in an hour and a half). However, be prepared to wait awhile if you want to print out all 2.9 billion base pairs of the human genome.