We will make the usual unrealistic independence assumptions. The stated assumptions of the problem are not enough. For example, we can perfectly well have equal probability of boy and girl on any birth, but a family stops having children once a boy is born. That would not affect overall sex distribution, but would have a strong effect on the makeup of families.
Imagine choosing a child at random. To make formulas simpler, let $$T=p_1+2p_2+3p_3+4p_4.$$
Then the probability the child is an only child is $\frac{p_1}{T}$, the probability the child comes from a $2$-child family is $\frac{2p_2}{T}$, the probability the child is from a $3$-child family is $\frac{3p_3}{T}$, and the probability the child is from a $4$-child family is $\frac{4p_4}{T}$.
By symmetry, the probability that a randomly chosen boy has a sister is the same as the probability that a randomly chosen child has a sibling of the opposite sex.
Things will be easier to visualize if we imagine that the chosen child is the oldest (it makes no difference).
We first calculate the probability that the chosen child does not have a sibling of the opposite sex. List the sequence of sexes of children, oldest first.
If the child is from a $1$-child family, the probability of having no sibling of the opposite sex is $1$.
If the child is from a $2$-child family, things get more complicated. There are $2$ possible sex orders once we know the sex of the oldest. For $1$ of these, there is no sibling of the opposite sex. So the probability the oldest has no sibling of the opposite sex is $1/2$.
If the oldest child is from a $3$-child family, there are $4$ equally likely sex distributions with the sex of the oldest child given. The oldest has no sibling of the opposite sex in only $1$ of these cases. So our probability is $1/4$.
Finally, there is a probability of only $1/8$ that the oldest child in a $4$-child family has no sibling of the opposite sex.
Add up. The probability that the oldest has no sibling of the opposite sex is
$$\frac{p_1}{T}+ \frac{1}{2}\cdot\frac{2p_2}{T}+\frac{1}{4}\cdot\frac{3p_3}{T}+\frac{1}{8}\cdot\frac{4p_4}{T}.$$
Finally, the probability that a randomly chosen child, say a boy, though it doesn't matter has a sibling of the opposite sex is obtained by subtracting the expression in $(\ast)$ from $1$.