Experiment / E1PUAWA85AAV-MPRA / in vivo MPRA

Xen1 pooled AAV SPRA screen in mouse lumbar dorsal horn

Combining Machine Learning and Multiplexed, In Situ Profiling to Engineer Cell Type and Behavioral Specificity

The full Xen1 pooled AAV library tested 27 candidate regulatory elements and controls, each represented by three unique synthetic barcodes, for a total of 81 designed constructs. After intraspinal delivery to mouse lumbar dorsal horn, barcode RNA was measured with a custom Xenium spatial transcriptomics panel and modeled with RESSCU to estimate cell-type and neuron-subtype specificity.

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

Spatial Parallel Reporter Assay (SPRA) used a non-integrating PHP.eB AAV pool carrying approximately 500-bp regulatory elements, each linked to three unique barcodes and a DL5/FAP reporter. Following intraspinal delivery to adult C57BL/6J mouse lumbar spinal cord and at least four weeks of expression, barcode RNA was measured at single-cell resolution with a custom Xenium panel. RESSCU negative-binomial models adjusted for mouse identity, cell area, spatial density of negative controls, and, for neuron-subtype models, hSyn/SV40 crosstalk.

Processed data

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

Visible columns (31 of 31)
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 · 31 definitions
barcode_model_id
RESSCU response-variable identifier for the barcode model, preserved exactly from the source workbook.
element_id
Library regulatory-element identifier parsed from the response-variable identifier, such as E14.
element_name
Human-readable enhancer, promoter, or control name from the supplemental library table.
element_type
Interpreted library category: candidate enhancer, promoter control, positive control, or negative control.
target_group
Cell type or neuronal target group associated with the element in the supplemental library design.
design_focus
Controlled design-focus label; these regulatory elements and controls are region-focused rather than allele-focused.
barcode_id
Individual synthetic barcode identifier parsed from the response-variable identifier.
model_level
RESSCU model level: cell_type or neuron_subtype.
term_class
Classification of the model term as intercept, covariate, or cell_identity.
effect_term
Exact RESSCU regression term from the source coefficient table.
estimate_log_count_effect
Natural-log coefficient estimate for the term; cell identity coefficients quantify attributable barcode-count effects after covariate adjustment.
std_error
Standard error of the coefficient estimate.
statistic
Regression test statistic for the coefficient.
p_value
Regression p-value for the coefficient.
conf_low
Lower confidence bound for the coefficient on the natural-log scale.
conf_high
Upper confidence bound for the coefficient on the natural-log scale.
exp_estimate_fold
Exponentiated coefficient, calculated during processing as a fold multiplier on the model's count scale; not an RNA/DNA ratio.
prop_positive
Fraction of cells positive for the modeled barcode from the corresponding model-fit statistics.
mcfadden_r2
McFadden pseudo-R2 for the corresponding cell-type model, when supplied.
auroc
Area under the receiver operating characteristic curve for the corresponding model.
auprc
Area under the precision-recall curve for the corresponding model.
mae
Mean absolute error for the corresponding neuron-subtype model; blank where not supplied for cell-type models.
null_mae
Mean absolute error for the corresponding null neuron-subtype model.
mae_nonzero
Mean absolute error restricted to nonzero observations for the corresponding neuron-subtype model.
null_mae_nonzero
Nonzero-observation mean absolute error for the corresponding null neuron-subtype model.
uniformity_p
DHARMa-style residual uniformity diagnostic p-value.
dispersion_p
DHARMa-style residual dispersion diagnostic p-value.
zero_inflation_p
DHARMa-style zero-inflation diagnostic p-value.
skewness
Skewness statistic for the barcode count distribution or model diagnostic, as supplied by the workbook.
paper_qc_pass
TRUE for coefficient/model rows retained in the author-supplied RESSCU workbook after the paper's QC; absent designed barcode IDs were not imputed.
source_sheet
Name of the source workbook sheet from which the coefficient row was taken.

Quality control

Author-reported QC retained Xenium cells with cell-label transfer confidence >0.5, excluded three small ambiguous neuron clusters, and retained 362,006 cells (63,610 neurons) across 3 Xen1 mice. Five of 81 synthetic barcode probes were excluded after high signal in two virus-free mice; RESSCU fit diagnostics were reported but were not used as per-barcode exclusion thresholds (4% zero-inflation p<0.05, 88% dispersion p<0.05, 48% non-uniformity p<0.05). The processed table retains only the 2,043 coefficient rows for the 75 barcode models supplied in the authors' RESSCU workbook; six designed barcode IDs have no model rows and are not imputed.

Curation notes

This package uses the authors' RESSCU summary outputs rather than a raw single-cell barcode count matrix, which was not included in the open supplementary package. The sequence PDF preserves the 500-bp element and barcode sequences in image-rendered Tables S11-S12. The paper reports five false-positive barcodes removed from the 81-probe panel, while media-4 contains 75 barcode model IDs; six designed IDs are absent from that workbook and no values were invented. Model estimates are natural-log count coefficients; exp_estimate_fold is calculated here and should not be interpreted as a conventional RNA/DNA MPRA activity ratio. The reference genome is null because the assay library combines macaque-, mouse-, human-derived, and synthetic/control elements rather than using one assembly.

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.