Prospective multi-centric study to analyze pre-transplant compatibility algorithm for live-related donor kidney transplant in Indian setting: the "Delhi approach"!

BACKGROUND: Since no single test is always accurate and sensitive, two or more tests are used to increase the precision of evaluation. Different algorithms have been proposed by centers in Leiden, Basel, Vienna and Minnesota, etc. With an intention to develop an optimal algorithm for India, we evaluated pre-transplant compatibility tests for live-donor kidney transplants. Three tests complement dependent cyto-toxicity cross-match (CDCXM), flow-cytometry cross-match (FCXM) and anti-HLA antibody screening (HAS) were performed and confirmed by the anti-HLA antibody identification (HAI) assay in a multi-centric trial (three transplant centers) in India. MATERIALS AND METHODS: All prospective recipients (and their potential donors) underwent low-resolution HLA typing as well as CDCXM, FCXM and HAS assays. In addition, HAI {single antigen bead assay; (SAB)} was done for all recipients to identify possible anti-HLA antibodies. In a virtual cross-match (VXM), antibody specificity was mapped to donor HLA type to determine donor-specific antibodies (DSA). Only patients without DSA were cleared for the transplant. Alternatively, patients with DSA were offered an exchange in the kidney paired donation (KPD) program. The screening results (CDCXM, FCXM, and HAS) were analyzed, individually as well as in combination of screening assays (CDCXM+HAS, CDCXM+FCXM, and FCXM+HAS) and the results were compared with those from the HAI test. RESULTS: Out of 100 patients, 69 were males and 31 were females; 85 recipients (85%) underwent a kidney transplant. The sensitivity of CDCXM was only 12.1% and the specificity of CDCXM was 100%; whereas the sensitivity of FCXM was 84.8% and the specificity of FCXM was 89.6%. The sensitivity and specificity of class I HAS was 88.2% and 84.3%, respectively. The sensitivity and specificity class II HAS was 88.0% and 80.0%, respectively. However, when both class I/II HAS were tested together the sensitivity increased to 97.0% and the specificity to 82.1%. Similarly, the sensitivity of combined FCXM+HAS had the sensitivity of 100% and the specificity of 76.1%; CDCXM+FCXM had the sensitivity of 84.8% and the specificity of 89.6% and CDCXM+HAS assays reached 97% with the specificity of 82.1%. CONCLUSIONS: Our results showed that the algorithm of FCXM with HAS produced the best sensitivity of 100%. The specificity of 76.1% indicate that the combined FCXM+HAS assays may detect up to 24.9% false positive results. We suggest that these false-positives may be easily resolved by performing the virtual crossmatch based on HAI (SAB) results. In our reflex testing algorithmic approach only 49% patients needed HAI (SAB). Finally, our results suggested that the CDCXM assay may be discontinued in pre-transplant workup owing to its very low sensitivity (12.1%).

Machine-OlF-Action: a unified framework for developing and interpreting machine-learning models for chemosensory research.

SUMMARY: Machine Learning-based techniques are emerging as state-of-the-art methods in chemoinformatics to selectively, effectively and speedily identify biologically relevant molecules from large databases. So far, a multitude of such techniques have been proposed, but unfortunately due to their sparse availability, and the dependency on high-end computational literacy, their wider adaptation faces challenges, at least in the context of G-Protein Coupled Receptors (GPCRs)-associated chemosensory research. Here, we report Machine-OlF-Action (MOA), a user-friendly, open-source computational framework, that utilizes user-supplied SMILES (simplified molecular input line entry system) of the chemicals, along with their activation status, to synthesize classification models. MOA integrates a number of popular chemical databases collectively harboring approximately 103 million chemical moieties. MOA also facilitates customized screening of user-supplied chemical datasets. A key feature of MOA is its ability to embed molecules based on the similarity of their local neighborhood, by utilizing a state-of-the-art model interpretability framework LIME. We demonstrate the utility of MOA in identifying previously unreported agonists for human and mouse olfactory receptors OR1A1 and MOR174-9 by leveraging the chemical features of their known agonists and non-agonists. In summary, here we develop an ML-powered software playground for performing supervisory learning tasks involving chemical compounds. AVAILABILITY AND IMPLEMENTATION: MOA is available for Windows, Mac and Linux operating systems. It's accessible at (https://ahuja-lab.in/). Source code, user manual, step-by-step guide and support is available at GitHub (https://github.com/the-ahuja-lab/Machine-Olf-Action). For results, reproducibility and hyperparameters, refer to Supplementary Notes. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.