I am new to working with Sanger traces and have been struggling with figuring out how to best assemble my sequences with them. I have been using lh3's seqtk trimfq for Phred based trimming. I essentially start at 0.05 error threshold and manually check if traces at the ends look good enough. If not, I decrease the threshold to 0.005 then to 0.001. Predictably I am getting a trade off of eliminating enough erroneous bases or having enough coverage to call a base. This wouldn't be a big deal if I was just working with a few short sequences, but I want to automate this on a larger scale without worrying about having to manually go back and check each assembly.

I am using CAP3 for assembly. I haven't figured out all the settings and am currently just using default. Ideally, from what I can tell I can tolerate a error threshold of 0.005 (this just means the sum of error tolerated over a subset? Not my Phred score cutoff right?) but I don't know if I can tell CAP3 to handle overlap specifically to ignore insertions?

Is there a standard or smarter approach? Thanks!