Is there a software/fun to fit a mixture of linear models?

Is there a software/fun to fit a mixture of linear models?

I have data set generated by y1 = ax + b + rand1 and y2 = px + q + rand2, where rand1 and rand2 are both Gaussian random variables with mean=zero and the same but unknown sigma. a, b, p and q are unknown parameters.

x is between [1, 2],, each data point has prior prob=0.7 belong to model I and prior prob=0.3 belong to model II.

Now, I need to find parameters a, b, p, and sigma through regression? Is there a matlab function or package for this task?

Thanks.