I would suggest that you should round off the fractional portion and store the whole number portion as 0111000.
8 bits is enough to store the numbers 0 to 255, or -128 to 127, or 0 to 127.5, or -64 to 63.5. or a whole lot of other options depending on how you decide to represent the number in binary.
However, you need to decide on a single representation, because otherwise there's no way to know that (for example) the number 11111110 is 254, -127, -63.5, or any of a plethora of different values.