A cross-sectional design is totally inappropriate for this type of question. The results are almost certainly driven by selection bias: i.e. depressed people behave differently. 

Without seeing changes from before and after the dietary change, you basically can’t learn anything. 

TLDR; nutrition research is almost always done poorly. 

Undisciplined

realBitcoinDog

HealthAndFitness

Mental health consequences of dietary restriction

0xbitcoiner

I'm not really the right one to answer that. But my general take is a varied diet is probably your best bet. Just keep in mind that everyone's unique, so what clicks for some might not for others.