Skip to main content

Table 1 Description of evaluated training sets

From: Current genomic deep learning models display decreased performance in cell type-specific accessible regions

Training set

Genomic regions

1

All genomic sequences

2

Sequences overlapping any ATAC peak and an equal number of non-peak sequences (1:1 peak to non-peak ratio)

3

Sequences overlapping any ATAC peak and an equal number of GC-matched non-peak sequences (1:1 peak to non-peak ratio)

4

Sequences overlapping any ATAC peak

5

Sequences overlapping non-ubiquitous ATAC peaks