I am trying to write a python script to update the header (only the first line) of some huge files, but as the new header is not necessary to be the same size (in bytes) as the original one, is there anyway I could change the header without touching the rest of the huge file? or I have to read through them all and write them back to file?

No, the only operations you can do on files without touching the whole file are truncation, replacement of same size, and appending.

You can, however, buffer relatively small parts of the file and write them after you've read all data currently residing in the new position, to avoid memory exhaustion. If speed is an issue, consider using mmap.