Nanopore sensing is now reshaping analytical proteomics with its simplicity, convenience, and high sensitivity. Determining the length of polyglutamine (polyQ) is crucial for the rapid screening of Huntington's disease. In this computational study, we present a cross-nanoslit detection approach to determine the polyQ length, where the nanoslit is carved within a two-dimensional (2D) in-plane heterostructure of graphene (GRA) and hexagonal boron nitride (hBN). We designed a heterostructure with an hBN strip embedded in the graphene sheet. With such a design, polyQ peptides can spontaneously and linearly stretch out on the hBN stripe. By tuning the strength of an external in-plane electric field, molecular transportation of polyQ peptides along the hBN stripe can be effectively regulated. Subsequent cross-nanoslit motion can be applied to record time-dependent electric signals. The signal features are then utilized to train the machine learning classification models. The machine-learning-assisted recognition enables accurate determination of the protein's length. This nanoslit-sensing method may offer theoretical guidance on 2D heterostructure design for the detection of polyQ peptide lengths and rapid screening of protein-related diseases.