bioRxiv Science⌕ Search

Biology subjects

Zhu, X.-W.

Publications and source records attributed to Zhu, X.-W..

2 recordsLinked to original sources
bioRxiv2019

Genotype Imputation and Reference Panel: A Systematic Evaluation

Here, 622 imputations were conducted with 394 customized reference panels for Han Chinese and European populations. Besides validating the fact that the imputation accuracy could always benefit from the increased panel size when the reference panel was population-specific, the results brought two new thoughts as follows. First, when the haplotype size of reference panel was fixed, the imputation accuracy of common and low-frequency variants (MAF>0.5%) decreased while the population-diversity of reference panel increased, but for rare variants (MAF<0.5%), a fraction of diversity (<20%) of panel could improve the imputation accuracy. Second, when the haplotype size of reference panel was increased with extra population-diverse samples, the imputation accuracy of common variants (MAF>5%) for European population could always benefit from the expanding sample size. But for Han Chinese population, the accuracy of all imputed variants reached the highest when reference panel contained a fraction of extra diverse sample (15%[~]21%). In addition, we evaluated the existing reference panels such as the HRC and 1000G Phase3 and CONVERGE. For European population, HRC was the best reference panel. For Han Chinese population, we proposed an optimum constituent ratio for the Han Chinese imputation if researchers would like to customize their own sequenced reference panel, but a high quality and large-scale Chinese reference panel was still needed. Our findings could be generalized to the other populations with conservative genome, a tool was provided to investigate other populations of interest (https://github.com/Abyss-bai/reference-panel-reconstruction).\n\nHighlights (Key points)O_LIA total of 394 reference panels were designed and customized by three strategies, and large-scale genotype imputations were performed with these panels for systematic evaluation in Han Chinese and European populations.\nC_LIO_LIThe accuracy of imputed variants reached the highest when reference panel contains a fraction of extra diverse sample (15%[~]21%) for Han Chinese population, if the haplotype size of the reference panel was increased with extra samples, which is the most common cases.\nC_LIO_LIThe imputation accuracy showed the different trends between Han Chinese and European populations. In a sense, the European genome may more diverse than Han Chinese genome by itself.\nC_LIO_LIExisting reference panels were not the best choice for Chinese imputation, a high quality and large-scale Chinese reference panel was still needed.\nC_LI

bioinformatics↗
bioRxiv2019

All-Assay-Max2 pQSAR: Activity predictions as accurate as 4-concentration IC50s for 8,558 Novartis assays

Profile-QSAR (pQSAR) is a massively multi-task, 2-step machine learning method with unprecedented scope, accuracy and applicability domain. In step one, a \"profile\" of conventional single-assay random forest regression (RFR) models are trained on a very large number of biochemical and cellular pIC50 assays using Morgan 2 sub-structural fingerprints as compound descriptors. In step two, a panel of PLS models are built using the profile of pIC50 predictions from those RFR models as compound descriptors. Hence the name. Previously described for a panel of 728 biochemical and cellular kinase assays, we have now built an enormous pQSAR from 11,805 diverse Novartis IC50 and EC50 assays. This large number of assays, and hence of compound descriptors for PLS, dictated reducing the profile by only including RFR models whose predictions correlate with the assay being modeled. The RFR and pQSAR models were evaluated with our \"realistically novel\" held-out test set whose median average similarity to the nearest training set member across the 11,805 assays was only 0.34, thus testing a realistically large applicability domain. For the 11,805 single-assay RFR models, the median correlation of prediction with experiment was only R2 ext=0.05, virtually random, and only 8% of the models achieved our standard success threshold of R2 ext=0.30. For pQSAR, the median correlation was R2 ext=0.53, comparable to 4-concentration experimental IC50s, and 72% of the models met our R2 ext>0.30 standard, totaling 8558 successful models. The successful models included assays from all of the 51 annotated target sub-classes, as well as 4196 phenotypic assays, indicating that pQSAR can be applied to virtually any disease area. Every month, all models are updated to include new measurements, and predictions are made for 5.5 million Novartis compounds, totaling 50 billion predictions. Common uses have included virtual screening, selectivity design, toxicity and promiscuity prediction, mechanism-of-action prediction, and others.

bioinformatics↗