Hattori-Kato, M. | bioRxiv Science

bioRxiv2018

Prediction of prostate cancer by deep learning with multilayer artificial neural network.

ObjectivesTo predict the rate of prostate cancer detection on prostate biopsy more accurately, the performance of deep learning utilizing a multilayer artificial neural network was investigated.\n\nMaterials and methodsA total of 334 patients who underwent multiparametric magnetic resonance imaging before ultrasonography-guided transrectal 12-core prostate biopsy were enrolled in the analysis. Twenty-two non-selected variables as well as selected ones by least absolute shrinkage and selection operator (Lasso) regression analysis and by stepwise logistic regression analysis were input into the constructed multilayer artificial neural network (ANN) programs. 232 patients were used as training cases of ANN programs, and the remaining 102 patients were for the test to output the probability of prostate cancer existence, accuracy of prostate cancer prediction, and area under the receiver operating characteristic (ROC) curve with the learned model.\n\nResultsWith any prostate cancer objective variable, Lasso and stepwise regression analyses selected 12 and 9 explanatory variables from 22, respectively. In common between them, age at biopsy, findings on digital rectal examination, findings in the peripheral zone on MRI diffusion-weighted imaging, and body mass index were positively influential variables, while numbers of previous prostatic biopsy and prostate volume were negatively influential. Using trained ANNs with multiple hidden layers, the accuracy of predicting any prostate cancer in test samples was about 5-10% higher compared with that with logistic regression analysis (LR). The AUCs with multilayer ANN were significantly larger on inputting variables that were selected by the stepwise logistic regression compared with the AUC with LR. The ANN had a higher net-benefit than LR between prostate cancer probability cut-off values of 0.38 and 0.6.\n\nConclusionANN accurately predicted prostate cancer without biopsy marginally better than LR. However, for clinical application, ANN performance may still need improvement.

bioinformatics↗

bioRxiv2017

Introduction of single nucleotide variations into the promoter region of the mouse kidney anion exchanger 1 gene downstream of a TATA box changes its transcriptional activity.

(Introduction) Distal renal tubular acidosis (dRTA) is characterized by an impairment of urine acidification and can be caused by variations in genes functioning in -intercalated cells such as anion exchanger 1 (AE1). AE1 encodes the Cl-/HCO3- exchanger in erythrocytes (eAE1) and -intercalated cells (kAE1). We previously reported that in human erythroid intron 3 containing the promoter region of human kAE1, a SNP (rs999716) 39 base pairs downstream of the TATA box showed a higher minor allele A frequency in incomplete dRTA patients and such a promoter region showed reduced activities, leading to the hypothesis that those with the A allele may express less kAE1, developing incomplete dRTA. Here, single nucleotide variations were introduced downstream of the TATA box in the murine erythroid intron 3 to investigate changes in the promoter activity for murine kAE1 mRNA. (Methods) The erythroid intron 3 of C57BL/6 was subcloned into the pGL4.17 reporter vector, leading to mu kAE1Pro-pGL4.17. Three types of G to A substitutions were introduced 33, 35, and 36 base pairs downstream of the TATA box in the murine kAE1 promoter region by inverse PCR (Var1, Var2, and Var3, respectively). The HEK 293 cells were transfected with vectors. After 24 hours, the firefly luciferase activity was determined. (Results and Discussion) The promoter activity of Var1, as well as that of Var2 to a lesser extent, was reduced compared with that of the wild-type. The introduction of variations such as Var1 and Var2 into the murine genome by genome editing may help to establish mouse models of incomplete dRTA.

genomics↗