Correlations conundrum A= a certain gene, B= observation, C= observation.
A is not correlated with B 
A is not correlated with C
However, 
B and C are correlated if A is present. 
B and C are not correlated if A is not present.
How can I explain this? 
Edited to add:
I'm not looking for a biological explanation, but more a statistical one. I don't quite understand how independent correlations are not present, but a combination is. Maybe it's my small N giving a type I or type II error? Apparently A is playing a role in the correlation between B and C, but not with B and C independently. 
Random example: gene A is not correlated with eating more apples (B) or eating more oranges (C). However, people who eat more apples also eat more oranges, but only if they have gene A. How?
 A: What exactly do you mean by "How can I explain this?"  Given the context, I'd say any satisfying or deep explanation will have to come from biology.
However, if you're asking "I don't see how this can ever be the case for any random variables (A, B, C) -- please show me an example," here's a simple one:
n_obs <- 10^5
df <- data.frame(A=sample(c(0, 1), n_obs, replace=TRUE, prob=c(0.25, 0.75)))
df$B <- runif(n_obs)
df$C <- ifelse(df$A, df$B, runif(n_obs))

cor(df$A, df$B)  # Essentially zero
cor(df$A, df$C)  # Essentially zero
with(subset(df, A == 1), cor(B, C))  # Given A==1, perfect correlation between B and C
with(subset(df, A == 0), cor(B, C))  # Given A==0, zero correlation between B and C

In this example $A \in \left\{0, 1\right\}$, and $B$ and $C$ are both uniform on $\left[0, 1\right]$.  Conditional on $A=0$ they are independent; conditional on $A=1$ they are perfectly correlated.
The key in this example is that the marginal distributions of B and C do not change with A. That still leaves room to play around with their joint distribution, e.g. by making them independent under one value of A and perfectly correlated under the other.

A: I'll give an intuitive example.


*

*Let $A$ be the event that the Philadelphia Eagles are playing the New York Giants.

*Let $B$ be the event that the Philadelphia Eagles win.

*Let $C$ be the event that the New York Giants lose.


Let's also assume the New York Giants and Philadelphia Eagles are evenly matched and both teams are about .500 teams against other opponents. Then we have:


*

*The event $A$ (Eagles playing Giants) is not correlated with either team winning or losing.

*If the Eagles play the Giants, event $B$ (Eagles winning) is perfectly correlated with event $C$ (Giants losing).

*If the Eagles are not playing the Giants, event $B$ and event $C$ are not correlated.

