.. _loading_data_int:

****************************************************
Loading the data and integration of multiple samples
****************************************************

To analyze multiple samples, select the **Integration** tab in the web
application. The full integration workflow is contained in this tab:
sample declaration, per-sample QC, normalization, integration method,
clustering, and the integrated visualization handoff.

Integration uses Seurat v5's
`IntegrateLayers <https://satijalab.org/seurat/reference/integratelayers>`_
framework. **RPCA** is the default method and **Harmony** is also
available from the same tab. Samples are declared directly in the UI —
pick the number of samples, type their names and paths inline, and set
the QC values for each on the same page.

.. figure:: images/v3/v3_integration.png
   :alt: Integration tab.
   :width: 100%
   :align: center

   Integration tab. Declare the number of samples and their paths in
   *Load Samples*, set per-sample QC in the middle column, and pick
   **RPCA (Seurat v5)** or **Harmony** in *Integration Method*.

Declaring the samples
=====================

In the *Load Samples* card:

#. Set **Number of samples** (1–24).
#. For each sample, fill in:

   - **Sample name** — any label you prefer. Use the same name for
     biological replicates that should be pooled in plots and analyses.
   - **10X data path** — relative or absolute path to the sample's
     directory.

Setting per-sample QC parameters
================================

In the *QC Parameters per sample* card, for each declared sample set:

- **Mito pattern** — regular expression matching mitochondrial gene IDs
  (default ``^MT-`` for human; use ``^mt-`` for mouse, ``^ATMG``
  for *Arabidopsis*, etc.).
- **Min genes** — minimum number of genes a cell must express to be
  retained.
- **Max genes** — maximum number of genes a cell can express
  (useful to filter suspected doublets).
- **Max mito %** — maximum percentage of mitochondrial transcripts
  allowed per cell.

Choosing the integration method
===============================

In the *Integration Method* card:

- **Method**: ``RPCA (Seurat v5)`` (default) or ``Harmony``.
- **Normalization method**: ``LogNormalize (faster)`` (default) or
  ``SCTransform v2``.
- **Number of PCs**: principal components to use during integration
  (default ``30``; range 5–100).

Click **Load & Integrate** to run the integration. The integrated
object becomes available to the clustering, DE, and visualization
modules below.

Saving the integrated object for reuse
======================================

Integration is the most time-consuming step in this tab. After it
completes, an *Integration Summary* card lets you download the
integrated Seurat object as an RDS file. To skip the integration step
later, load that RDS in the **Single Sample** tab using *Existing
Seurat RDS* — the clustering and downstream steps will pick up from
there.