Gene expression measured in post-mortem human hippocampus tissue by Blalock et al. (2004), GEO accession GSE1297. The workbooks contain 22,283 probe sets for 31 samples, one per person, in four groups: Control, Incipient AD, Moderate and Severe.
It traces those workbooks to the 22,283 × 31 design matrix used by the Workflow page Alzheimer's gene co-expression: one row per probe set, one column per person. How the tissue was measured is taken from the study’s description; the preparation steps are recomputed here from the workbooks, which already contain processed expression values rather than scanner images.
Blalock EM, Geddes JW, Chen KC, Porter NM, Markesbery WR, Landfield PW
- DocumentedHippocampal tissue → microarray
GEO GSE1297 identifies the Affymetrix Human Genome U133A platform (GPL96).
- DocumentedArray signals → MAS5
Blalock et al. describe MAS5 processing in the study methods.
- Not recordedStudy output → workbook export
The exact export history of the contributed workbooks is not recorded.
- HeldProbe and severity workbooks
5 workbooks retain expression values and probe identifiers.
- ReproducedGlobal scaling within each group
All 22,283 probes retained; each severity block is scaled with its own single mean and standard deviation.
How to read the labels: Documented — described by the study. Held — files this page has. Reproduced — recomputed here from those files. Not recorded — the files do not show how the step was done.
What the instrument measures
The study measured RNA from post-mortem hippocampal CA1 tissue using a separate Affymetrix HG-U133A microarray for each subject. This is a tissue-level expression assay, so each sample combines signals from its constituent cells.
Schematic of the measurement steps described by the study.
Published acquisition protocol
Documented Blalock et al. (2004), Methods; GEO GSE1297 / GPL96. Protocol quantities below are from this source.
| Stage | Protocol |
|---|---|
| Specimen preparation | University of Kentucky brain-bank hippocampal sections were frozen in liquid nitrogen, stored at −80 °C, and dissected to isolate CA1. |
| RNA and labelling | The published protocol used 8 µg total RNA per sample to produce biotin-labelled cRNA; 20 µg cRNA was applied per array. |
| Assay platform | Affymetrix Human Genome U133A (HG-U133A), GEO platform GPL96; one array per subject. |
| Signal processing | The study used Microarray Suite 5 (MAS5) for relative expression signals and detection statistics. |
| Study screening | The paper required detection on at least four chips and blanked within-group outliers beyond 2 SD. This workflow does not apply these filters. |
The study processed its arrays with MAS5. This page starts from the contributed workbooks and keeps all their probe rows.
One acquisition unit: one subject’s tissue sample on one array
Each measurement unit is one tissue sample on one microarray. Each expression column is one sample; each row is a probe set.
What the fields mean
| Field | Unit | Meaning |
|---|---|---|
| ID_REF | Identifier | Affymetrix probe-set identifier; the stable row key. |
| IDENTIFIER | Annotation | Gene name supplied with a probe; several probes can share a gene name. |
| Control sample accession columns | Expression value | One processed expression reading per probe and sample. |
| Disease workbook numeric columns | Expression value | One sample per column; positional rows follow the identifier workbook. |
In the files: 22,283 probe sets × 31 sample readings, one sample per person.
| File | Bytes | Rows | Columns |
|---|---|---|---|
| dataset.xlsx | 1,874,068 | 22,283 | 11 |
| dataset1.xlsx | 1,423,226 | 22,283 | 9 |
| dataset2.xlsx | 1,137,355 | 22,283 | 7 |
| dataset3.xlsx | 1,284,836 | 22,283 | 8 |
| dataset4.xlsx | 1,143,428 | 22,283 | 7 |
A file, opened
First workbook rows, including the identifier header and Control sample identifiers.
ID_REF IDENTIFIER GSM21215 GSM21217 GSM21218 GSM21219 GSM21220 GSM21221 GSM21226 GSM21231 GSM21232 1007_s_at MIR4640 2735.0 3746.0 3317.0 5689.0 4192.1 3048.6 4723.8 4662.9 5801.2 1053_at RFC2 42.7 26.1 138.9 77.6 86.2 129.4 48.4 111.3 34.3 117_at HSPA6 161.3 153.6 381.2 248.8 894.1 98.2 233.7 86.6 328.0 121_at PAX8 1272.3 979.6 917.4 1291.9 1176.9 1242.9 1659.7 1398.0 1631.5 1255_g_at GUCA1A 312.7 444.6 97.5 206.4 189.8 197.5 137.1 120.2 142.0 1294_at MIR5193 320.8 319.9 297.0 470.8 240.2 285.1 542.2 526.9 707.5 1316_at THRA 135.9 307.3 172.6 166.8 103.6 158.3 190.1 162.7 194.6 1320_at PTPN21 12.7 11.3 15.1 23.6 32.8 17.4 25.1 20.5 27.9
What the values look like
Median expression across all probes for each sample, in workbook order. Colours show each sample’s group.
All values in this summary
Samples per group
| Group | Samples |
|---|---|
| Control | 9 |
| Incipient AD | 7 |
| Moderate | 8 |
| Severe | 7 |
- Expression is already processed: workbook values are not scanner pixels, and the files do not name an intensity unit.
- Probe is not gene: ID_REF identifies a probe set; multiple probe sets can name the same gene.
- Row order binds the files: disease workbooks contain numeric values without identifiers. Their positional correspondence follows the submitted loader and notebook.
- No time axis: sample position is workbook order, not an acquisition date. Only the Control identifier table retains sample accessions.
These steps follow the project’s notebook and stop at the matrix that goes into dimensionality reduction. Embeddings, clustering and their interpretation are on the Workflow page.
| Step | Evidence | Operation | Result |
|---|---|---|---|
| 0a | Documented | Study microarrays → MAS5 expression signals (Blalock et al., Methods) | Published study processing |
| 0b | Not recorded | Study expression output → contributed workbooks | Held processed values |
| 1 | Reproduced | Read every probe row; require finite numeric values | 22,283 rows; 0 missing values |
| 2 | Reproduced | Join severity blocks in Control → Incipient → Moderate → Severe order | 22,283 × 31 |
| 3 | Reproduced | For each severity block, subtract its single global mean and divide by its single global standard deviation | 22,283 × 31 |
| 4 | Reproduced | Keep probe and gene identifiers and the group labels beside the matrix, not as features | 22,283 probe IDs; 4 column groups |
Scaling uses one mean and SD for an entire severity block, not one per probe or sample. It therefore retains differences in probe amplitude.
| Block | Global mean | Global SD |
|---|---|---|
| Control | 684.003070 | 1728.233019 |
| Incipient AD | 678.600066 | 1697.548792 |
| Moderate | 694.506213 | 1851.736226 |
| Severe | 686.458978 | 1854.014168 |
What remains
Counts by group
| Group | Input rows | Stored rows |
|---|---|---|
| Control | 22283 | 22283 |
| Incipient AD | 22283 | 22283 |
| Moderate | 22283 | 22283 |
| Severe | 22283 | 22283 |
Follow one probe
1 · workbook values
2 · the stored row
Example values at every shown step
No probe is dropped: every reading is a valid number. The examples span low, middle and high expression and include both anchor probes, PLD2 and TPP2.
Checks against the supplied Workflow
| Check | Result |
|---|---|
| Combined input digest | Matches |
| Control workbook equals identifier table | Matches |
| Probe and sample counts | Matches |
One row is one probe set; the 31 columns are individual samples, grouped by severity. Rows preserve workbook probe order. The diverging scale shows signed, standardised expression. Column colours identify severity groups.
Prepared values span -0.39946 to 63.50089. Heatmap display only: colour saturates at ±3; 11,174 values exceed that display range. The heatmap uses colour steps of 0.02362, and screen pixels compress rows (and, for wide matrices, columns). The selected-row chart and table show the values without that colour rounding.
Open this section to load the embedded matrix.
Hover a row or use the row selector.
Every feature of the selected row
| Feature columns | Unit | Meaning |
|---|---|---|
| 0–30 | Standard deviations | Sample expression after global within-severity scaling |
Labels kept beside the matrix
| Label | Meaning |
|---|---|
| Probe ID | ID_REF: probe set identifier |
| Gene name | IDENTIFIER: annotation; not necessarily unique |
| Severity and sample ID | Column annotations; disease sample IDs are not retained in the numeric workbooks |
- All 22,283 probe rows are retained. Probe-level co-expression is not evidence of a direct gene interaction.
- The study describes MAS5 processing, but the workbooks do not record how their values were exported from it.
- Disease sample IDs and acquisition dates are not present in the numeric workbooks; row correspondence follows the supplied notebook.
- 58 gene-annotation cells contain spreadsheet error text and are shown as “No readable annotation”; the probe IDs remain the row identifiers.
- A single global scale per severity block does not normalise individual probe profiles.
Blalock et al. (2004). Incipient Alzheimer’s disease: Microarray correlation analyses reveal major transcriptional and tumor suppressor responses. PNAS. DOI 10.1073/pnas.0308512100. GEO GSE1297 / GPL96.
The Workflow page Alzheimer's gene co-expression uses these profiles for t-SNE mapping and K-Means clustering; its Insight page presents the results.