I see a couple of issues.
First:
ser.read() is only going to return 1 byte at a time.
If you specify a count
ser.read(5)
it will read 5 bytes (less if timeout occurrs before 5 bytes arrive.)
If you know that your input is always properly terminated with EOL characters,
better way is to use
ser.readline()
That will continue to read characters until an EOL is received.
Second:
Even if you get ser.read() or ser.readline() to return multiple bytes,
since you are iterating over the return value, you will
still be handling it one byte at a time.
Get rid of the
for line in ser.read():
and just say:
line = ser.readline()