---
id: "claim:normalization-for-sampled-count-data-2026-06-22-biorxiv-v4:dc699880719c6e117c84"
kind: "claim"
title: "Changing the pseudocount can alter PC loadings, k-nearest-neighbor graphs, and Leiden clusters."
description: ""
url: "https://sina.bio/claim/changing-the-pseudocount-can-alter-pc-loadings-k-nearest-neighbor-graph-d109a152"
---

# Changing the pseudocount can alter PC loadings, k-nearest-neighbor graphs, and Leiden clusters.

## Content

Changing the pseudocount can alter PC loadings, k-nearest-neighbor graphs, and Leiden clusters.

## Metadata

```json
{
  "claim_position": 122,
  "source_publication_id": "version:normalization-for-sampled-count-data--2026-06-22-biorxiv-v4",
  "source_work_id": "work:normalization-for-sampled-count-data"
}
```

## Connections (2)

### Normalization for sampled count data

- ID: `version:normalization-for-sampled-count-data--2026-06-22-biorxiv-v4`
- Kind: `publication`
- Direction: `incoming`
- URL: <https://sina.bio/publication/normalization-for-sampled-count-data>

This publication contains this grounded claim.

> The practical consequence is not limited to the value of a technical-variance diagnostic: changing the pseudocount can change PC loadings, the 𝑘-NN graph built from the PCs, and ultimately Leiden clusters (Fig. 2f and Supplementary Note Fig. 3).

Source: 2026-06-22-biorxiv-v4.pdf — <https://sina.bio/source/2026-06-22-biorxiv-v4-pdf>

### A shifted-log scale near one million generally fit benchmark data better than 10,000, and estimating overdispersion to s

- ID: `result:normalization-for-sampled-count-data-2026-06-22-biorxiv-v4:7ae79ddff0f33f83e129`
- Kind: `result`
- Direction: `incoming`
- URL: <https://sina.bio/result/a-shifted-log-scale-near-one-million-generally-fit-benchmark-data-bette-28dbb7fc>

This claim supports this result.