Advanced stochastic sampling techniques are used to probe the conguration space of self-avoiding exible rings subject to spatial connement. The sampled ring conforma- tions are next processed by algorithms that exactly classify knots in rings that are not excessively conned. The physical parameters of the rings, i.e. contour length, bending rigidity and steric hindrance, are properly chosen to mimic the properties of DNA of the P4 bacteriophage, while the radius of the smallest sphere used to conne the model rings is about 2.5 times larger than the P4 capsid radius. By comparing the computed knot spectrum with the P4 one we establish that the consideredmodel correctly predicts an increase of the occurrence of chiral knots with progressive connement but does not account for the bias of torus versus twist knots observed in P4 experiments.
