Overcalled A's in my 454 data

[entropy]

I've assembled 780,000 reads of 454 sequence with the Roche gsAssembler program and I've noticed that there are a huge number of overcalled A's. Most of them seem to be associated with runs of A's, which is no surprise, but the fact that its all one base, and almost always an insertion. Its almost like the settings for acquiring that channel on the instrument were off. It looks like the program has collapsed the gaps caused by these insertions in the final contigs, but I'm wondering if this is inhibiting some of the contigs from being linked.

Is there anyway to "turn down the A's", or "re-call" the bases in the data?

Thanks