Abstract
The output of icosahedral symmetry-adapted functions (ISAF) reconstruction algorithm is a virus capsid, where the density function of every outside point is theoretically zero. However, original ISAF algorithm takes reconstructed map as a solid sphere and density function of every point in the whole sphere is calculated. The density functions outside virus capsid may not be always zero, which may sometimes produce noise. In this study, Fourier shell correlation (FSC) is used for proving that density function calculation of original ISAF algorithm can reduce reconstruction precision. Subsequently, the range of virus capsid is determined with radial density function curve and the original ISAF algorithm is improved by calculating density function inside virus capsid. The experimental results on Cryo-electron microscopy (Cryo-EM) data of cytoplasmic polyhedrosis virus (CPV) indicate that the improved ISAF algorithm can generate more details of virus capsid and achieve a higher resolution.