but InSNP can find even rare indels because it looks at each individual sequence.
When you click Find Indels, it looks for any positions on any sequence where the best:total peak area ratio average is 0.9 or more for the forty bases before the position and 0.8 or less for the forty bases after it. When it finds such a sequence, it presents you with a new window:
In this window, the top trace is the putative indel.
The second trace is from a sequence without the indel.
The third trace is the difference between the indel trace and half of the clean sequence. This represents what the signal from the strand containing the indel might look like if you could sequence it by itself.
The fourth trace is half of the clean sequence, up to the position of the indel.
The fifth trace is half of the clean sequence, after the indel.
If this really is an indel, you should be able to subtract traces four and five from trace three and wind up with virtually no signal (no peaks) at all. That difference is represented by the bottom trace. To make it right, you must determine the length of the indel; looking at the sequence after the indel position in trace three is a good starting point. For our example here, the indel is a one base deletion, so entering -1 into the indel length field smooths out the bottom trace. This is a good indel, so we click Accept Indel.
Once you have approved an indel, the program looks for other sequences with that indel, and trims all sequences that have it.