There is no such thing as a recommended dosage for an unknown drug. Every drug has its own dosage based on it toxicity and pharmacokinetics profile. A non toxic drug at 0.5 mg/kg dosage is better than a potent drug at 0.05 mg/kg with a narrow therapeutic window. There are ways to translate in vitro cytotoxicity data to predict in vivo efficacy. The first thing you need to know is to determine the acute toxicity profile of your drug in vivo, say in mice or rats that you want to experiment with. Determine the Lethal dose (LD)50 and you will know the dose that you can not go beyond. Your goal is to show efficacy at dosage far away from the dose that can possible cause some form of toxicity in the test animals.
I usually use 0.1, 1, 10, 100 and 1000 mg/L for the preliminary tests to see in which range I get the results form multiple repeats. After that I narrow down this window to 2 orders of magnitude and do a precise measurement. There is no appropriate range for a dose range. If our compound is functional at 100 mg/L but it kills healthy cells at 110 mg/L it is useless for further studies but if it is functional at 100 but affects healthy cells at 500 mg/L it would be a candidate for further studies.
As far as I know, there is no recommended dose to start with due to a lot of biological factors/interferences along with the potency of the drug itself. Generally, you need to do a preliminary toxicity study using the minimum number of mice to identify the LD50. Thereafter, you can start your experiment looking for the best effective therapeutic dose which should be less than the LD50 you identified.
As per my experience from in vivo study, you need to conduct toxicity study first, then have to optimize the treatment dose according to in vitro study dose. you may look for appropriate method to calculate from in vitro dose to in vivo. I agreed with Dr. Abdual Rahman as well. You may follow OECD guidelines section 420 for toxicity study or up and down method section 425. Thank you wassalam.