We designed two strategies to integrate the outputs of these tools, enabling researchers to select appropriate polyA site sets based on their specific needs.
The main pipeline is divided into two parts:
- Position-based Method: Calculate the distance between genomic coordinates of sites from different sources, grouping sites with distances below a specified threshold into clusters, with each cluster representing a true signal.
- Similarity-based Method: Perform unsupervised clustering of sites from different sources based on expression similarity. After optimizing the number of clustering centers, the resulting clusters represent the integrated results.
- Install Nextflow (version 23.04.3.5875)
- Create conda environment (yaml in the conda_env folder)
This pipeline consists of three directories:
- bin: R scripts
- modules: Nextflow processes
- conda_env: conda environment
The main.nf is the pipeline to integrate polyA sites. Besides, we provided two workflows to get the input:
- apatools.nf: run Sierra, polyApipe, and SCAPE tools
- preprocess.nf: include site annotation, expression normalization and conversion from count matrix to bed format
Before running the integration pipeline, users need to run apatools.nf and preprocess.nf to get the input.
The usage of apatools.nf is as follows:
nextflow run apatools.nf --sample_name ${sample_name} --method ${apa_method} --data_dir ${input} --core_num 8 --job_id ${SLURM_JOB_ID} --barcode_file ${barcode_file} --reference_file ${reference_file} --genome_version ${genome_version} -profile conda
The ${input} is the output folder of cellranger or spaceranger which includes bam file and bam.bai file. The ${genome_version} needs to be consistent with ${reference_file}. If the genome is mm10, then the reference should be refdata-gex-mm10-2020-A/genes/genes.gtf. If the genome is GRCm39, then the reference should be refdata-gex-GRCm39-2024-A/genes/genes.gtf. Both references can be downloaded from [Cell Ranger website] (https://www.10xgenomics.com/support/cn/software/cell-ranger/downloads#reference-downloads).
The usage of preprocess.nf is as follows:
nextflow run preprocess.nf --sample_name ${sample_name} --method ${apa_method} --apa_input ${apa_input} --core_num 8 --barcode_file ${barcode_file} --reference_file ${reference_file} --genome_version ${genome_version} --gene_dir ${gene_dir} --annotation_dir ${annotation_dir} -profile conda
The ${apa_input} is the output folder of apatools.nf which includes all tools' outputs. The ${annotation_dir} is the folder of annotation of sites (one of outputs of the workflow). The ${gene_dir} is the folder to store gene-level matrix which will be used to normalize apa counts.
The usage of the integration pipeline is as follows:
- Run one condition
nextflow run main.nf --sample_name ${sample_name} --apa_input ${apa_input} --core_num 8 --barcode_file ${barcode_file} --distance_method "cosine" --cluster_method "kmeans-wss" --distance "50" --metatools_output ${metatools_output} -profile conda
- Run multiple conditions
nextflow run main.nf --sample_name ${sample_name} --apa_input ${apa_input} --core_num 8 --barcode_file ${barcode_file} --distance_method "[spearman, cosine, Jaccard, pearson, none, norm, euclidean]" --cluster_method "[hdbscan, hclust, pam-wss, pam-asw, kmeans-wss, kmeans-asw]" --distance "[0, 3000, 100]" --metatools_output ${metatools_output} -profile conda -with-report ${sample_name}_report.html -with-timeline ${sample_name}_timeline.html
Zhao, Q., & Rattray, M. (2025). metaAPA: a tool for integration of PolyA site predictions from single-cell and spatial transcriptomics