What you need is (1) pharmacology doses in animal models - (2) PK data from at least three animal species from the following - mouse, rat, dog or monkey. (3) GLP tox data including pathology with NOAEL (No Observed Adverse Effect Level) doses with toxicokinetic analysis.
From animal PK data allometric scaling can be conducted to predict or calculate human PK. However, sometimes it is not accurate because of species differences in metabolism and drug transport. In such cases, some in vitro data on metabolism and transport across species is helpful in correcting the PK data for species differences and predicting accurate clinical PK.
Pharmacology data for efficacy is helpful to predict what might be clinical efficacious dose(s). A PK/PD scientist can do modeling and simulations to get an idea of clinical efficacious doses and an appropriate efficacious plasma concentration.
Toxicology data Typically pivotal 1 month in at least one rodent and one large animal is important along with Toxicokinetic data. These data can help determine what might be the appropriate safety margin between efficacious concentrations and NOAEL concentrations. These NOAELs can be used to calculate appropriate human equivalent dose (HED). According to FDA guidance on toxicokinetics one has to use the most sensitive toxicology species for calculation of safe human dose but again it depends on species differences in other metabolism, absorption etc. (Guidance for Industry) A safety factor of 10 is applied for the most sensitive species to calculate the safe starting dose in humans. I think you should read the guidance: Estimating the Maximum Safe Starting Dose in Initial Clinical Trials for Therapeutics in Adult Healthy Volunteers, July 2005 (http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/ucm078932.pdf)