A total of 254,113 particles were extracted from micrographs, scaled by a factor of 0.25, and subjected to 3D classification into 8 classes in RELION. The particle coordinates corresponding to the four highest resolution classes that showed density for all lid components were re-centered, based on the translations determined from 3D classification. These coordinates were used to extract 139,561 unbinned particles for 3D refinement with RELION, which yielded a reconstruction of 4.4 Å. After correcting for particle motion and electron beam damage (particle polishing), the resolution was improved to 4.1 Å. A 3D mask surrounding the most structurally stable regions of the map (the PCI domains, the MPN heterodimer, and the helical bundle) was generated, and used in an alignment-free 3D classification of the data into 3 classes. The 109,396 particles contributing to the highest resolution 3D class were used for further 3D refinement in RELION without applying a mask, yielding a 3.6 Å structure. Continued refinement of these alignment parameters using the same 3D mask that was applied earlier improved the resolution of the central regions of the structure by 0.1 Å.