Experiment / E138PWX0QIntegrated lentiMPRA

K562 large-scale lentiMPRA benchmark dataset

Designing DNA With Tunable Regulatory Activity Using Discrete Diffusion

The paper benchmarks D3 on a public K562 large-scale lentiMPRA library from Agarwal et al. (2025), consisting of genomic candidate enhancers, promoters, and controls measured in three biological replicates. This package contains the corresponding ENCODE processed element quantifications reused by the paper; it does not represent the paper’s separate 2,760-element synthetic-validation library.

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

Lentivirus-based genomic integration of 230-bp candidate regulatory elements upstream of a minimal promoter with the barcode in the 5' UTR; DNA and RNA barcode counts were processed with MPRAflow for three biological replicates. The raw table contains ENCODE-normalized element-level DNA and RNA counts, RNA/DNA ratios, log2 activity scores, and observed barcode counts.

Processed data

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

Visible columns (27 of 27)
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 · 27 definitions
element_id
Exact ENCODE element identifier/name; reverse names preserve the source suffix.
element_class
Category inferred from the element name: candidate enhancer, promoter, dinucleotide-shuffled control, or other/control.
chromosome
GRCh38 chromosome from the associated BED file; blank when no coordinate was available.
start_0based
GRCh38 BED zero-based start coordinate.
end_0based
GRCh38 BED exclusive end coordinate.
strand
GRCh38 BED strand (+ or -).
orientation
Orientation derived from the BED strand: forward or reverse.
replicate_count_passing
Number of biological replicates passing the package QC filters.
min_barcode_count
Minimum observed barcode count across the three passing replicates.
dna_count_rep1
ENCODE normalized element DNA count for biological replicate 1.
rna_count_rep1
ENCODE normalized element RNA count for biological replicate 1.
ratio_rep1
ENCODE normalized RNA/DNA ratio for biological replicate 1.
log2_activity_rep1
ENCODE log2 RNA/DNA activity score for biological replicate 1.
n_barcodes_rep1
Observed barcode count (n_obs_bc) for biological replicate 1.
dna_count_rep2
ENCODE normalized element DNA count for biological replicate 2.
rna_count_rep2
ENCODE normalized element RNA count for biological replicate 2.
ratio_rep2
ENCODE normalized RNA/DNA ratio for biological replicate 2.
log2_activity_rep2
ENCODE log2 RNA/DNA activity score for biological replicate 2.
n_barcodes_rep2
Observed barcode count (n_obs_bc) for biological replicate 2.
dna_count_rep3
ENCODE normalized element DNA count for biological replicate 3.
rna_count_rep3
ENCODE normalized element RNA count for biological replicate 3.
ratio_rep3
ENCODE normalized RNA/DNA ratio for biological replicate 3.
log2_activity_rep3
ENCODE log2 RNA/DNA activity score for biological replicate 3.
n_barcodes_rep3
Observed barcode count (n_obs_bc) for biological replicate 3.
log2_activity_mean
Arithmetic mean of the three replicate log2 activity scores.
log2_activity_sd
Sample standard deviation of the three replicate log2 activity scores.
qc_pass
TRUE for rows retained after the package QC filters.

Quality control

The ENCODE MPRAflow combined table was filtered to remove the no_BC record; require all three biological replicates; require at least 10 observed barcodes (n_obs_bc) per replicate, positive normalized DNA and RNA counts, and finite ratio/log2 values. This retained 224,555 of 230,932 non-no_BC element IDs (all 3 replicates), and all retained rows have qc_pass=TRUE. GRCh38 BED coordinates were joined where available; missing coordinates did not cause exclusion.

Curation notes

The open v3 article reports a separate wet-lab K562 validation library of 2,760 elements and 2,534 retained elements, but its MPRAflow validation table is not included in the open supplement or linked Zenodo record. This package therefore represents the public ENCODE K562 large-scale lentiMPRA benchmark dataset explicitly reused by the paper (Agarwal et al. 2025), not the study’s new synthetic-vs-genomic validation results. The table retains exact ENCODE element names; reverse-orientation names include the _Reversed: suffix. dna_count/rna_count are the normalized count fields supplied by ENCODE, not raw FASTQ read counts. Aggregate activity SD is the sample standard deviation across the three replicates.

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.