The Specific Aim of this SBIR Phase I project is to build a technology for discovery of monoclonal antibody (mAb) drug candidates for pancreatic cancer from rare primary B cells. Pancreatic cancer has proven resilient to drugs that target oncology's "usual suspects", including proteins related to angiogenesis, proliferation, and metastasis (Mackenzie &McCollum, 2009). mAbs are therapeutic agents of choice for oncology (Gura, 2002;Elder, 2011). Traditionally, mAb discovery programs identify a lead target antigen and screen antibody libraries for affinity using either phage display or mouse hybridomas. As an alternative, some academic and industry groups have been using human B cell hybridomas (Lang et al., 1990;Yu et al., 2008) to mine natural repertoires for candidate therapeutic antibodies. B cell hybridomas are technically challenging and inefficient, but natural antibodies have several advantages, including low immunogenicity, in vivo affinity maturation, and excellent expressability in mammalian cells (Hoet et al., 2005;Beerli &Rader, 2010). Next-generation sequencing (NGS) deep repertoire sequencing methods may help overcome the technical problems with B cell repertoire mining. Canonical NGS methods are used to deep sequence immunoglobulin (Ig) heavy or light monomers produced by clonal B or plasma cells. However, conventional NGS methods miss the single-cell context of paired Ig, requiring guesswork to pair Ig monomers and affinity screen (Reddy et al., 2010). GigaGen uses high-tech genomics to generate, for the first time, DNA libraries with native Ig pairing from millions o single human B cells. We address many of the shortcomings with existing methodologies through the following innovations: (i) deep droplet digital single cell genomics;(ii) detection of affinity maturation in vivo;(iii) comparisons among tissues, time points, and patients;and (iv) integration with protein display. In Phase I, we will adapt our technology specifically for pancreatic cancer mAb discovery. First, in order to capture rare antibodies, we will increase the cell throughput of our microfluidic device at least tenfold. Second, we will develop protocols for disaggregation of tumor infiltrating B cells (TIL-Bs) from pancreatic tumors. We will accomplish the Specific Aim by performing the following tasks: (i) Validate a novel microfluidic chip geometry for >10x higher cell throughput;(ii) Test methods for disaggregation and sorting of TIL-Bs from pancreatic tumors;and (iii) Sequence pancreatic cancer patient peripheral blood and TIL-Bs to discover lineages undergoing affinity maturation. We will be successful if we achieve the following metrics: (i) Generate droplets at 60-200?m in diameter, <5% droplet merging, multiple-cell encapsulation rates of <3%, and cell throughput of at least 400Hz across at least 20 experiments;(ii) Mis-linking of heavy and light chain Ig should not differ significanty between libraries generated on the old and new chips (two-proportion z-test;?=0.05;power=0.8);(iii) Recover at least 100,000 unique B cell clones from each of 6 pancreatic TIL-B libraries;and (iv) Discover >10 Ig sequences undergoing affinity maturation and shared between TIL-Bs and peripheral blood from the same patient, or undergoing convergent evolution. Phase I will generate a curated list of candidate antibodies with strong therapeutic potential. In Phase II, we will bring our candidate mAbs through standard preclinical development (Zhao et al., 2009). We expect this stage of development to cost $1-2m and last 18-24 months. By the end of Phase II, we hope to have sufficient data for clinical development of a novel lead mAb, either with a biopharma partner or by raising venture capital.