I have isolated Genomic DNA of different moths species. I sequenced using gene specific primer. I observed noise in sequence data sheet first 20 base pairs. We need to trim the sequence.
The explanation for why it happens I don't know, but I find there always to be rubbish sequence at the beginning of a sanger read. It's usually the first 10-50 bases.
The low quality of the first 10-50 bp is as a result of primer binding. Are your primers ~20 bp in length? Subsequent trimming is necessary to ensure you only use sequence with confidently assigned base calls to minimize the possibility of unreliable and incorrect database hits.
If you are using Sanger sequencing you should always design your primers in such a way as to allow for the noise in the first part of the read. Alternatively, if you sequence is under ~600bp you can bypass this problem by performing bi-directional sequencing on single products. The reverse sequencing will provide information on the 'noisy region' in the first bit of your forward sequencing.
If you require the sequence from the beginning to the end I would suggest cloning the PCR product into a vector and use vector specific primers to sequence it. Thus the noise will only be present in the vector part.
The answer is in your question itself. Obviously if there are noise in your sequence data, you must remove it. You may also need to trim your sequences later while performing alignment prior to phylogenetic tree construction.
Trimming is necessary because normally noise may find between the primer binding site1-50 bp, some sequences it may find upto 80 bp. Such cases you can do bi-directional sequencing to avoid this problem. Forward sequence always gives you the better sequence result while trimming the noise region.