Comparative ChIP-STARR-seq: NSC-derived HIST and TF libraries in embryonic stem cells

The same pooled NSC-derived HIST (H3K27ac/H3K4me1) and TF (SOX2/YY1) ChIP-STARR-seq plasmid libraries were transfected into H9 embryonic stem cells to measure comparative episomal activity across the 148,198-region scaffold. The table reports normalized STARR RNA/plasmid DNA activity and the study's comparative ESC activity ranks, with the corresponding NSC measurement retained for comparison.

Processed tables are specific to each experiment. Column names, units, measurements, and table structure are not standardized across the database. Check this experiment’s column definitions and quality-control notes before comparing or combining data.

Organism: Human
Taxonomy ID: NCBITaxon:9606
Biosample: CVCL:9773
Reference genome: GRCh38
Design focus: Region-focused
Region of interest: Not reported / not applicable

Perturbation & assay details

5 uM ROCK inhibitor (Y27632) during suspension transfection

Comparative ChIP-STARR-seq is an episomal self-transcribing reporter assay using the same NSC-derived ChIP-enriched plasmid libraries as the NSC condition; target enrichment was performed by ChIP rather than capture hybridization. Genomic fragments were cloned downstream of a minimal-promoter-driven GFP and upstream of the polyadenylation signal; five independent suspension transfections per plasmid pool were pooled after 24 hours. RNA and plasmid DNA sequencing were used to calculate activity. The paper reports four indexed RNA replicates per HIST or TF pool and two plasmid-library DNA replicates per pool.

Quality control

Study QC: NSC ChIP was performed in duplicate and ChIP-seq replicates showed Pearson correlation >0.88; independent indexed STARR RNA technical replicates showed Pearson correlation >0.96 and were merged for downstream analysis. Reads were adapter-trimmed, aligned to GRCh38/hg38 with Bowtie2 using --very-sensitive, and only properly and uniquely mapped reads with MAPQ >=30 were retained; ChIP duplicates were removed with Picard. Study scaffold activity was computed after removing low-coverage regions with fewer than 20 reads in at least two samples, using normalized RNA/plasmid read ratios averaged across replicates. Package QC retained one row per unique published NCRE location, requiring finite NSC and ESC log2(avg.RPP+1) values, parseable chromosome coordinates, and a 500-1000 bp scaffold length. All 148,198 unique scaffolds passed these package filters.

Curation notes

The source Table S2 contains repeated rows for a scaffold when it has multiple target-gene/HPO annotations. The processed table collapses exact duplicate coordinates and combines unique annotations with pipes; activity values and ESC categories are taken from the ESC sheet, while the NSC activity/category fields are retained for comparison. The paper removed a small number of patch sequences before CNN training, but Table S2 does not expose a patch flag; this package represents the published measured activity atlas and does not silently remove those rows. Individual low-throughput validation constructs and BRAIN-MAGNET nucleotide contribution scores were not substituted for the atlas activity table. Biosample CVCL:9773 is the Cellosaurus WA09/H9 embryonic stem cell entry.

Row
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50

Comparative ChIP-STARR-seq: NSC-derived HIST and TF libraries in embryonic stem cells

Processed data

Quality control

Curation notes

Cite OpenMPRA

Row
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50

Row
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50

Row
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50