Which is already sequenced - in this case you can include in your study sequences from GenBank and make your results compatible. 

But it give different result when aligned and tree mad with reported sequences and different with original

When you remake the tree with only the sequences used in the original source you are referencing, do you get the same result? Are you using the same model and parameters as used to generate the original reference? Assuming replicating the original gives the same answer, then the changes come from adding more data, which is unlikely to actually be a problem. As long as the extra data is not bogus, the addition of data often changes our results and interpretations somewhat.