The conventional design with only one primary study population (an overall population) has recently been challenged), particularly when the disease (e.g. cancer) is heterogeneous due to observable clinical characteristics and/or unobservable underlying biologic/genomic characteristics. In recent oncology clinical trials, biomarker subpopulations (biomarker +/-) have become increasingly important for drug development in tailored therapies to fulfill regulatory commitments. It is also desirable to optimize the study design to have a more enhanced claim in which multiple objectives can be met. In this work, we propose to extend the classical group sequential design set-up to the above setting and develop the methods and calculation with less conservative bounds and, therefore, a smaller required sample size or greater power.