Treatment of type 2 diabetes (T2D) involves a variety of medications, related to the stage of T2D progression. In silico trials can be helpful in this context, speeding up the drug development process and supporting treatment optimization. We recently proposed a T2D simulator (T2DS), consisting of a model of the glucose-insulin system and an in silico population describing glucose-insulin dynamics in T2D subjects. Both T2DS model and virtual population have been developed based on early-stage T2D data studied with sophisticated experimental techniques. This limits the T2DS domain of validity to this specific sub-population of T2D. Conversely, in order to provide the most suitable and effective testing platform, the T2DS should be equipped with an additional virtual cohort well reflects the characteristics of the population object of study. In principle, this would require further complex experiments for each population under study, a time-demanding and expensive task. Alternatively, we propose a method for tuning the T2DS to any desired T2D target population, e.g. insulin-naïve (i.e., not experienced with insulin) patients, without the need to resort to complex and expensive clinical studies. To illustrate the method, we provide a case study aiming at extending the T2DS to insulin-naïve Caucasian individuals with T2D. First, the method consists on identifying the T2DS model on available literature data of the target population. The estimated parameters are then used as new reference for generating a new cohort of virtual subjects that are more representative of the Caucasian T2D population. Then, a model of basal insulin degludec (iDeg) is also incorporated into the T2DS in order to enable insulin therapy. The targeted T2DS is finally validated by simulating iDeg therapy initiation and comparing the simulated outcomes with the clinical ones. As result, we show that the simulated distributions of fasting plasma glucose and iDeg dose reproduce clinical data, meaning that the tuned T2DS is representative of Caucasian T2D subjects, and thus it can effectively support therapy optimization for the target population. The methodology described here can be extended to other stages of T2D, allowing an extensive in silico testing phase of different treatments before human trials.