OncoCyrix

OncoCyrix is a modular, production-ready Scanpy pipeline for processing and analyzing a single 10x Genomics single-cell RNA-seq sample.

The pipeline is optimized for human cancer datasets, but works for any standard 10x scRNA-seq run.

---

Key Capabilities

• 10x Genomics matrix ingestion (MTX + barcodes + features)
• Gene ID normalization (Ensembl → HGNC symbols)
• Quality control filtering
  • Mitochondrial percentage
  • UMI counts
  • Genes per cell
• Normalization and log1p transformation
• Highly variable gene (HVG) selection
• PCA, UMAP, and t-SNE embeddings
• Leiden clustering
• Cell type annotation using CellTypist
• Cell-type-specific marker discovery
• Pathway enrichment analysis
  • GO (BP, MF, CC)
  • KEGG
  • Reactome
  • WikiPathways

Final biological summary

Cell Types → DEGs → Marker Genes → Pathways

---

Project Structure

singlecell_pipeline/
├── config_cli.py          # CLI and global configuration
├── loader_10x.py          # 10x data loading
├── gene_names.py          # Gene ID normalization
├── group_de.py            # Differential expression analysis
├── markers.py             # Marker gene detection
├── pathway_enrichment.py  # Enrichment analysis and deduplication
├── summary_ct_deg.py      # Integrated summaries
├── pipeline.py            # Scanpy orchestration
└── main_single.py         # Pipeline entry point

---

Features in Detail

1. 10x Data Loading

• Automatically detects:
  • matrix.mtx / matrix.mtx.gz
  • barcodes.tsv / barcodes.tsv.gz
  • features.tsv or genes.tsv
• Efficient sparse matrix handling

2. Gene Name Normalization

• Detects Ensembl gene IDs
• Maps to HGNC gene symbols using mygene.info
• Ensures unique and consistent gene names

3. Quality Control & Filtering

• Computes:
  • pct_counts_mt
  • n_genes_by_counts
  • total_counts
• Filters:
  • <200 or >6000 genes per cell

  • 15% mitochondrial reads

  • Genes expressed in fewer than 3 cells

4. Normalization & HVG Selection

• Library size normalization
• Log1p transformation
• HVG selection (Seurat v3 flavor)

5. Dimensionality Reduction

• PCA (50 components)
• UMAP
• t-SNE (enabled for datasets with fewer than 50k cells)

6. Clustering

• Leiden clustering (default resolution = 0.5)
• Cluster-level visualizations

7. Cell Type Annotation

• Metadata-based annotation or
• Machine-learning-based prediction using CellTypist
• PCA / UMAP / t-SNE plots colored by cell type

8. Marker Gene Detection

• Global marker genes
• Cell-type-specific markers
• Rank plots, heatmaps, and dotplots

9. Pathway Enrichment

• Enrichment via gseapy / Enrichr
• Supported databases:
  • GO Biological Process
  • GO Molecular Function
  • GO Cellular Component
  • KEGG
  • Reactome
  • WikiPathways
• Semantic deduplication using MiniLM + FAISS

10. Integrated Biological Summary

Automatically links:

• Cell types
• DEGs
• Marker genes
• Enriched pathways

---

Usage

Run the pipeline on a single 10x dataset:

scpipeline   --single-10x-dir "/path/to/10x_folder"   --single-sample-label TumorA   --single-group-label LUNG_CANCER

All results are saved to:

<10x_folder>/SC_RESULTS/

---

Outputs Generated

• Quality control plots
• Highly variable gene tables
• PCA / UMAP / t-SNE embeddings
• Clustering results
• Cell type annotations
• Marker gene tables
• Pathway enrichment results
• Integrated summary tables

---

Docker Usage

Docker image available on Docker Hub:

docker pull sheryar09/scpipeline:latest

Run:

docker run --rm   -v /path/to/10x_folder:/data   sheryar09/scpipeline:latest   --single-10x-dir /data   --single-sample-label TumorA   --single-group-label LUNG_CANCER

---

Intended Use Cases

• Cancer single-cell RNA-seq analysis
• Tumor microenvironment profiling
• Biomarker discovery
• Translational and preclinical studies
• ML-based cell type prediction

---

Version: 1.0
Author: Sheryar Malik
Project Name: OncoCyrix