For Research Use Only (RUO)

PanTek Analysis Lab

Upload a research-use proteomics matrix, generate QC figures, and export publication-ready PNG drafts directly in the browser.

第一版工具不會上傳或儲存您的資料。所有解析、圖表與匯出都在瀏覽器端完成。

PanTek Analysis Lab v0.5 — frontend-only research prototype

Upload CSV / TSV Back to Service Page

Upload Panel

支援 CSV / TSV wide-format protein intensity matrix。大型資料可在瀏覽器端執行，但 cohort-scale 專案仍建議使用正式 workflow。

Input mode (?)

Raw intensities (vendor output — Skyline, MaxQuant, DIA-NN, etc.) Log2-transformed (already log2 before upload) Normalized (median / quantile / VSN; centered distribution)

Normalization mode (?)

None (identity — no normalization applied) Median-center (per-sample, log2 space) Quantile (per-sample rank-average, log2 space)

Choose Matrix File

Required: one matrix file with protein annotations and numeric sample columns.

Drop file here, load demo data, download an example matrix, or open the data format guide.

View Data Format Guide

Empty state: upload a CSV/TSV file or load the synthetic demo dataset to begin.

File Summary

上傳後會顯示檔案摘要、分隔符號、列欄數與前 5 列 preview。

Rows-

Columns-

Samples-

Delimiter-

Preview
No file loaded.

Sample Metadata (Optional)

選用：上傳第二個 CSV / TSV 檔案，提供每個 sample 的補充資訊（例如收案組別、批次、處理時間）。Metadata 僅以 in-memory 形式暫存；重新上傳 matrix 會清除。本版本支援以 metadata 欄驅動 render-only chart stratification；exports 預設將 grouping 標籤 pseudonymize 為 G1 / G2 / … 以保護第三方識別（可在下方明確 opt-out）。

Choose Metadata File

Optional: one metadata file with a sample-ID column matching the matrix sample columns.

Drop file here or use Choose Metadata File above. Analysis still runs without it.

Empty state: upload a metadata CSV/TSV to map sample-level attributes, or skip if not needed.

Data Requirements

Wide-format proteomics matrix

至少需要 protein ID 欄位與 2 個 numeric sample intensity 欄位。缺失值、NA、NaN、null 與 0 會在 presence/missingness 中視為 missing；圖表轉換使用 log2(x + 1)。

Supported columns

工具會自動偵測 protein_id、gene_symbol、protein_name 與 sample intensity columns。若命名不同，可在 mapping UI 重新指定。

protein_id,gene_symbol,protein_name,Sample_01_NEAT,Sample_02_NEAT,Sample_03_SEER,Sample_04_SEER
P02768,ALB,Albumin,123456,120345,34567,33210
P02787,TF,Serotransferrin,23456,23890,19876,18765

Processing rules

Presence matrix is computed before imputation. PCA uses proteins valid in at least 50% of samples and row median imputation only for visualization. Imputation never changes protein ID counts.

CSV / TSV log2(x + 1) Row median PCA imputation Browser memory only

Data Format Guide

PanTek Analysis Lab supports one wide-format matrix where rows are proteins and columns are samples. Required columns are a protein identifier plus at least two numeric sample intensity columns. Recommended annotation columns are gene_symbol and protein_name.

Example header: protein_id,gene_symbol,protein_name,Sample_01_NEAT,Sample_02_NEAT,Sample_03_SEER,...

Intensity values must be numeric. Empty values, 0, NA, NaN, null, and non-finite values are treated as missing for presence and missingness calculations.

Uploaded files are parsed locally in your browser and are not transmitted to PanTek servers in this demo version.

Pre-analysis Validation Report

Validate before chart generation

The validation report checks required columns, sample intensity quality, missingness, replicate sufficiency, baseline availability, marker availability and module readiness before charts are generated.

No validation report yet.

Upload a matrix or load the demo dataset to inspect pass, warning and error states.

Column Mapping

Review detected columns before analysis

自動偵測結果會先填入下方欄位；您可以手動調整 annotation 欄、sample columns、condition label 與 baseline condition。

No matrix loaded yet.

Upload CSV/TSV or load the demo dataset to enable mapping controls.

Protein ID column

Gene symbol column

Protein name column

Sample intensity columns

Numeric columns are selected by default. Edit condition labels, include/exclude sample columns, then apply mapping to regenerate all charts.

Include	Sample column name	Detected condition	Editable condition	Numeric valid count	Missing rate

Changing baseline automatically recalculates the point-biserial / signed CV ratio module when valid results are available.

Sample QC Dashboard

File-level and sample-level readiness

Dashboard summarizes protein depth, missingness, total intensity, albumin ratio, top-10 dominance, replicate sufficiency and sparse matrix warnings.

Dashboard is waiting for data.

Load a dataset to generate sample QC summary.

Chart Gallery

Plasma proteome QC and benchmark figures

Every generated figure has a PNG export button and a traceable CSV table. Skipped modules report the reason instead of crashing.

No charts generated yet.

Run analysis after loading a matrix.

Overview QC PCA Quantification Markers Downloads

Protein ID Count by Sample / Condition

Detected proteins per sample, where valid intensity is non-missing and greater than zero.

Method note: counts use the raw presence matrix before any imputation; missing, zero, NA, NaN and null are not counted as detected.

Missingness by Sample

Missing or zero intensity fraction by sample.

Method note: missingness is computed from raw values before imputation. Protein valid-rate distribution summarizes detected fraction across included samples.

Method Overlap

Shared proteins across all conditions, condition-specific proteins, and partial overlaps.

Method note: condition sets are built from proteins detected in at least one sample per condition. Pairwise overlap is Jaccard-style count overlap, not statistical differential analysis.

PCA Plot

log2(x + 1), proteins valid in at least 50% samples, row median imputation for PCA only.

Method note: PCA uses log2(x + 1), proteins valid in at least 50% samples, and row median imputation only for visualization.

Albumin Ratio vs Protein ID Count

Albumin relative intensity against detected protein depth.

Method note: albumin is detected by gene_symbol ALB or protein name containing albumin. Correlation is Pearson on sample-level albumin ratio and protein ID count.

CV Distribution by Condition

Raw-intensity coefficient of variation per protein and condition.

Method note: CV is calculated from raw observed intensities within each condition; conditions with fewer than two samples are skipped for replicate statistics.

Point-biserial Correlation and Signed CV Ratio

Baseline vs enriched condition consistency using r_pb and signed CV ratio.

Comparison

Method note: r_pb compares baseline vs enriched condition using log2 intensity. Signed CV ratio uses raw-intensity CV; positive values indicate higher CV after enrichment and negative values indicate lower CV.

Sample Composition Marker Scores

Platelet, erythrocyte / hemolysis, and coagulation marker median log2 intensity by sample, plus depth relationship.

Show in dashboard / report figures

Toggle controls visual display + visual exports (PNG / DOCX / HTML report). Data tables remain available in CSV / JSON exports regardless of this toggle.

Method note: marker scores are median log2 intensities of built-in platelet, erythrocyte / hemolysis and coagulation marker genes found in the uploaded matrix.

Low-abundance Marker Coverage

Detected count and coverage rate for cytokine-like, growth-factor-like, hormone-like and chemokine/interleukin-like panels.

Show in dashboard / report figures

Toggle controls visual display + visual exports (PNG / DOCX / HTML report). Data tables remain available in CSV / JSON exports regardless of this toggle.

Method note: marker coverage uses fixed built-in research panels; denominator is the panel size and detected markers are those present in any sample of a condition.

Ranking Plot

Condition-level median log2 intensity ranking with ALB and marker genes highlighted.

Condition

Method note: ranking uses condition-level median raw intensity transformed to log2 for display. Highlights include ALB, contamination markers and low-abundance marker panels.

Dynamic Range Summary

High-abundance protein dominance and log2 intensity spread by sample and condition.

Method note: top 10 / top 50 contribution, log2 IQR and log2 range are computed from observed raw intensities per sample to flag high-abundance dominance.

Feature Matrix Readiness Score

Sample-level Pass / Warning / Review flags for downstream feature matrix readiness.

Method note: readiness is a descriptive heuristic using missingness, protein depth, contamination score and PCA distance when available. It is exploratory and RUO only.

Before / After Normalization (per-sample)

Per-sample log-intensity distribution before and after the selected normalization. Research-use-only data view, rendered in-browser only — it is not included in any figure, table, or report export.

Method note: each box summarizes one sample's positive observed intensities (null, non-finite and non-positive cells excluded) on a log axis. Before = retained raw matrix; After = the normalized matrix for the active mode. This QC view is intentionally not captured into PNG / CSV / ZIP / HTML / DOCX / JSON exports.

Download Center

Export figures, tables, reports, and summary files

所有輸出都由目前瀏覽器記憶體中的分析結果生成。此 demo 不會將資料寫入 browser storage 或傳送到 PanTek servers。

Report Studio

Generate a Word-ready research report

PanTek Multiomic Analysis Lab — Research-use Proteomics QC and Enrichment Benchmark Report. Word export is generated entirely in browser memory.

Report type

Project title

Dataset label

Analyst / team name

Notes (optional)

Do not enter patient names, direct identifiers, or personally identifiable clinical information. No API key is required or used.

Future API extension: this frontend can export a compact manuscript_payload.json for a backend service to generate manuscript-style narratives or DOCX reports with Gemini or other LLM APIs. API keys must never be placed in frontend code.

Run analysis first, then preview or export the report package.

Report Preview Outline

Preview the sections, figures, tables, and skipped modules before generating the Word report.

No report outline yet.

For Research Use Only.
This report studio supports exploratory data visualization and early research scoping. It does not provide clinical diagnosis, medical advice, treatment decisions, IVD testing, or regulated bioinformatics interpretation.

ZIP export packages generated PNG snapshots, CSV tables, analysis_summary.json, manuscript_payload.json, report outline, report assets, and a standalone HTML report from the current browser-memory analysis.

Limitations / RUO Disclaimer

Exploratory visualization, not regulated interpretation

For Research Use Only. This tool is intended for exploratory data visualization and early research scoping. It does not provide clinical diagnosis, medical advice, treatment decisions, IVD testing, or regulated bioinformatics interpretation.

Browser-only analysis is suitable for small to moderate matrices. Formal cohort-scale workflows should use validated backend processing, auditable run manifests, and controlled project governance.

PanTek Analysis Lab

Upload Panel

File Summary

Sample Metadata (Optional)

Join preview

Wide-format proteomics matrix

Supported columns

Processing rules

Validate before chart generation

Pass

Warning

Error

Module Status

Review detected columns before analysis

Sample intensity columns

File-level and sample-level readiness

Plasma proteome QC and benchmark figures

Protein ID Count by Sample / Condition

Missingness by Sample

Method Overlap

PCA Plot

Albumin Ratio vs Protein ID Count

CV Distribution by Condition

Point-biserial Correlation and Signed CV Ratio

Sample Composition Marker Scores

Low-abundance Marker Coverage

Ranking Plot

Dynamic Range Summary

Feature Matrix Readiness Score

Before / After Normalization (per-sample)

Export figures, tables, reports, and summary files

Generate a Word-ready research report

Report Preview Outline

Exploratory visualization, not regulated interpretation

Upload Panel

File Summary

Sample Metadata (Optional)

Join preview

Wide-format proteomics matrix

Supported columns

Processing rules

Validate before chart generation

Pass

Warning

Error

Module Status

Review detected columns before analysis

Sample intensity columns

Condition labels

File-level and sample-level readiness

Plasma proteome QC and benchmark figures

Protein ID Count by Sample / Condition

Missingness by Sample

Method Overlap

PCA Plot

Albumin Ratio vs Protein ID Count

CV Distribution by Condition

Point-biserial Correlation and Signed CV Ratio

Sample Composition Marker Scores

Low-abundance Marker Coverage

Ranking Plot

Dynamic Range Summary

Feature Matrix Readiness Score

Before / After Normalization (per-sample)

Export figures, tables, reports, and summary files

Generate a Word-ready research report

Report Preview Outline

Exploratory visualization, not regulated interpretation