Experiment / E6NN660EFWhole-Genome STARR-seq (WHG-STARR-seq)

Five-genome pooled whole-genome STARR-seq in K562 cells

The architecture of allele-specific regulatory variant effects across five human genomes

A pooled whole-genome STARR-seq library was generated from sheared genomic DNA from five Yoruba individuals in the 1000 Genomes Project, combined, and assayed in K562 cells. Three independent transfections/output RNA libraries and the matched pooled input DNA library were used to call regulatory elements and estimate allele-specific activity.

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.

Perturbation & assay details

Basal / Untreated

Episomal STARR-seq: approximately 500 bp human genomic DNA fragments were cloned into the 3′ UTR of a synthetic reporter, allowing inserts to regulate their own transcription. Five Yoruba 1000 Genomes libraries were pooled; the average fragment size was approximately 450 bp, with approximately 2 billion unique fragments and approximately 300× human-genome coverage. The pooled library was transfected into K562 cells approximately 150 million cells per replicate, harvested 6 hours later, and sequenced in triplicate. The processed table uses the authors’ BIRD posterior variant-effect supplement; the package also retains the compact ENCODE CRADLE replicated element-call file.

Processed data

50 rows per page. Click a cell to inspect its full value.

Visible columns (13 of 13)
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

Filters apply to this table only. The CSV download contains the complete processed table; filtered rows are available through the API.

Column dictionary · 13 definitions
variant_id
Source variant identifier in the form chrom:position:reference_allele:alternate_allele
chrom
GRCh38 chromosome from the source variant identifier
position
Source genomic variant position; coordinate convention is retained as provided by the authors
reference_allele
Reference allele sequence from the source variant identifier
alternate_allele
Alternate allele sequence from the source variant identifier
variant_type
Derived class: SNV, insertion, deletion, or MNV based on reference and alternate allele lengths
theta
BIRD posterior median allelic regulatory-effect odds ratio; values above 1 indicate alternate-allele overrepresentation in reporter RNA relative to input DNA
log2_effect_size
Base-2 logarithm of theta; positive values indicate increased alternate-allele reporter activity and negative values indicate decreased activity
lower_ci
Lower bound of the authors’ 95% symmetric credible interval for theta
upper_ci
Upper bound of the authors’ 95% symmetric credible interval for theta
posterior_effect_probability
BIRD posterior probability that the variant has a regulatory effect (source field Preg)
high_confidence_effect
Derived flag true when posterior_effect_probability is greater than 0.9, matching the paper’s high-confidence count of 5,393 variants
credible_interval_excludes_null
Derived flag true when the 95% credible interval excludes the null odds ratio theta=1

Quality control

The study tested variants with more than 5 reads in the pooled input library and in each replicate output library. Replicate activity was highly correlated over the downstream union set (0.81 ≤ r ≤ 0.88), and de novo regulatory regions were called at 5% irreproducible discovery rate. The processed table was additionally checked for the expected five source fields, valid GRCh38 chromosome/position and allele structure, positive finite theta and credible-interval values, and posterior probabilities in [0,1]. All 187,493 source records passed these checks; no rows were filtered.

Curation notes

This is one pooled K562 whole-genome STARR-seq condition with three biological/output replicates, not five separate cell-line experiments. The paper reports approximately 200,000 tested variant effects; the authors’ MiniSTARR repository supplies the compact 187,493-row BIRD variant-effect table used here. The raw_data directory also retains Tables 1–5, the ENCODE CRADLE-corrected replicated element calls, ENCODE/GEO metadata and sample readmes, and the open preprint. Raw FASTQ/BAM files and the three large signal bigWigs were excluded because raw reads are unnecessary for this package and the bigWigs are redundant with the compact calls and exceed the practical package target.

Cite OpenMPRA

Cite the OpenMPRA database. Include your access date because the collection changes over time.

Please also cite the source studies when using their data.