Table 1 The performance of the cell-specific D-miRT models on the 10% test data.

From: A two-stream convolutional neural network for microRNA transcription start site feature integration and identification

Cell line

# Pos

# Neg

Accuracy

Precision

Recall

F1-score

A549

7645

742

0.9577 (0.8660)

0.9629 (0.8753)

0.9511 (0.8503)

0.9570 (0.8626)

GM12878

9935

943

0.9329 (0.8081)

0.9473 (0.7971)

0.9215 (0.8129)

0.9342 (0.8049)

GM12878*

9935

943

0.9243 (0.8491)

0.9420 (0.8608)

0.9219 (0.8490)

0.9293 (0.8509)

Hela-S3

9968

960

0.9410 (0.8302)

0.9422 (0.8254)

0.9379 (0.8320)

0.9400 (0.8287)

HepG2

10345

986

0.9418 (0.8381)

0.9509 (0.8316)

0.9332 (0.8444)

0.9420 (0.8379)

hESC

9120

873

0.9450 (0.8425)

0.9511 (0.8620)

0.9447 (0.8229)

0.9479 (0.8420)

K562

8019

762

0.9479 (0.8722)

0.9310 (0.8722)

0.9534 (0.8712)

0.9421 (0.8717)

K562*

8019

762

0.9388 (0.9079)

0.9386 (0.9160)

0.9385 (0.9079)

0.9385 (0.9096)

MCF-7

9259

901

0.9470 (0.8531)

0.9505 (0.8293)

0.9341 (0.8973)

0.9422 (0.8619)

  1. The numbers in parenthesis are those without threshold values. The #Pos and #Neg are the number of positive samples and negative samples used to test the trained cell-specific models. The two rows with “*” next to cell line names are from the two-layer models.
Back to article page