If you call, it's 7 into a pot that will be 20 on the flop. Ignoring rake, that means you need to win about 35% of the time for continuing to be better than folding.
On a basic level, you need to find out which hands have at least 35% equity vs villain's range, and then take away a few that don't realize equity very well OOP (e.g. offsuit aces) and add a few that over-realize their equity (e.g. suited connectors). You can do this with Equilab's 'Hand Range Calculator' function.
If, for example, villain's 10% 3-bet range looked something like: 77+, ATs+, A5s, KJs+, QJs, JTs, AJo+, KQo
then the only hands with 35%+ equity are 22+, A2s+, K9s+, QTs+, JTs, T9s, A8o+, A5o, KQo, which is 19% of hands (254 combos). But from experience and previous study, I would fold A9o and A8o and A5o, and replace them with 54s+, and probably QJo and KJo, K9s and Q9s. I might add some suited gappers like J9s, T8s, 97s as well.
Assuming I 4-bet 99+/AJ+ vs the 10% 3-bettor, I would end up flatting 88-22, ATs-A2s, K9s+, Q9s+, J9s+, T9s, 98s, 87s, 76s, 65s, 54s, ATo, KJo+, QJo (14%). And if I was opening about 45% of the time, I'm folding about 55% of the time to the 3-bet, which seems fairly reasonable to me, as playing bloated pots OOP without the lead (especially vs a larger than pot-sized 3-bet) is quite difficult.
Against a more aggro player, I'd increase the 4-betting for value range, and also put in some 4-bet bluffs with hands that probably don't make money as calls. I also might do a fair bit of limping in. You can't get 3-bet if you limp.