26Therapeutic treatments for late-onset Alzheimer's disease (LOAD) are hindered by an incomplete 27 understanding of the temporal molecular changes that lead to disease onset and progression. Here, we 28 evaluate the ability of manifold learning to develop a molecular model for the unobserved temporal 29 disease progression from RNA-Seq data collected from human postmortem brain samples collected 30 within the ROS/MAP and Mayo Clinic RNA-Seq studies of the AMP-AD consortium. This approach 31 defines a cross-sectional ordering across samples based on their relative similarity in RNA-Seq profiles 32 and uses this information to define an estimate of molecular disease stage -or disease progression 33 pseudotime -for each sample. This transcriptional estimate of disease progression is strongly concordant 34 with burden of tau pathology (Braak score, P = 1.0x10 -5 ), amyloid pathology (CERAD score, P = 1.8x10 -35 5 ), and cognitive diagnosis (P = 3.5x10 -7 ) of LOAD. Further, the disease progression estimate 36 recapitulates known changes in cell type abundance and impact of genes that harbor known AD risk loci. 37Samples estimated to reside early in disease progression were enriched for control and early stage AD 38 cases, and demonstrated changes in basic cellular functions. Samples estimated to reside late in disease 39 progression were enriched for late-stage AD cases, and demonstrated changes in known disease processes 40 including neuroinflammation and amyloid pathology. We also identified a set of control samples with 41 late-stage estimated disease progression who also showed compensatory changes in genes involved in 42 affected pathways are protein trafficking, splicing, regulation of apoptosis, and prevention of amyloid 43 cleavage. In summary, we present a disease specific method for ordering patients based on their LOAD 44 disease progression from CNS transcriptomic data. 45