Spatial Transcriptomics Pipeline
The ST Pipeline contains the tools and scripts needed to process and analyze the raw files generated with the Spatial Transcriptomics technology in FASTQ format to generate datasets for down-stream analysis. The ST pipeline can also be used to process single cell data as long as a file with barcodes identifying each cell is provided (same template as the files in the folder "ids").
The ST Pipeline has been optimized for speed, robustness and it is very easy to use with many parameters to adjust all the settings. The ST Pipeline is fully parallel and has constant memory use.
The following files/parameters are required :
- FASTQ files (Read 1 containing the spatial information and the UMI and read 2 containing the genomic sequence)
- A genome index generated with STAR
- An annotation file in GTF or GFF3 format (optional when using a transcriptome)
- The file containing the barcodes and array coordinates (look at the folder "ids" and chose the correct one). Basically this file contains 3 columns (BARCODE, X and Y), so if you provide this file with barcodes identinfying cells (for example), the ST pipeline can be used for single cell data. This file is also optional if the data is not barcode (for example RNA-Seq data).
- A name for the dataset
The ST pipeline has multiple parameters mostly related to trimming, mapping and annotation but generally the default values are good enough. You can see a full description of the parameters typing "st_pipeline_run.py --help" after you have installed the ST pipeline.
The input FASTQ files can be given in gzip/bzip format as well.
Basically what the ST pipeline does is :
- Quality trimming (read 1 and read 2) :
- Remove low quality bases
- Sanity check (reads same length, reads order, etc..)
- Check quality UMI
- Remove artifacts (PolyT, PolyA, PolyG, PolyN and PolyC) of user defined length
- Check for AT and GC content
- Discard reads with a minimum number of bases of that failed any of the checks above
- Contamimant filter e.x. rRNA genome (Optional)
- Mapping with STAR (only read 2)
- Demultiplexing with Taggd (only read 1)
- Keep reads (read 2) that contain a valid barcode and are correctly mapped
- Annotate the reads with htseq-count (slightly modified version)
- Group annotated reads by barcode(spot position), gene and genomic location (with an offset) to get a read count
- In the grouping/counting only unique molecules (UMIs) are kept.
You can see a graphical more detailed description of the workflow in the documents workflow.pdf and workflow_extended.pdf
The output will be a matrix of counts (genes as columns, spots as rows), a BED file containing the transcripts (Read name, coordinate, gene, etc..), and a JSON file with useful stats. The ST pipeline will also output a log file with useful information.
We recommend you install a virtual environment like Pyenv or Anaconda before you install the pipeline. The ST Pipeline works with python 2.7.
You can install the ST Pipeline using PyPy:
pip install stpipeline
Alternatively, you can build the ST Pipeline yourself:
First clone the repository
git clone <stpipeline repository>
or download a tar/zip from the releases section and unzip it
Access the cloned ST Pipeline folder or the folder where the tar/zip file has been decompressed.
To install the pipeline type then
python setup.py build python setup.py install
To run a test type (you need internet connection to run the tests)
python setup.py test
To see the different options type
An example run would be
st_pipeline_run.py --ids ids_file.txt --ref-map path_to_index --log-file log_file.txt --output-folder /home/me/results --ref-annotation annotation_file.gtf file1.fastq file2.fastq
If you used an Ensembl annotation file and you would like change the ouput file so it contains gene Ids/names instead of Ensembl ids. You can use this tool that comes with the ST Pipeline
convertEnsemblToNames.py --annotation path_to_annotation_file --output st_data_updated.tsv st_data.tsv
Merge demultiplexed FASTQ files
If you used different indexes to sequence and need to merge the files you can use the script merge_fastq.py
merge_fastq.py --run-path path_to_run_folder --out-path path_to_output --identifiers S1 S2 S3 S4
Where identifiers will be strings that identify each demultiplexed sample.
Filter out genes by gene type
If you want to remove from the dataset (matrix in TSV) genes corresponding to certain gene types (For instance to keep only protein_coding). You can do so with the script filter_gene_type_matrix.py
filter_gene_type_matrix.py --counts-matrix stdata.tsv --gene-types-keep protein-coding --outfile new_stdata.tsv --annotation path_to_annotation_file
You may include the parameter --ensembl-ids if your gene names are represented as gene ids instead.
Remove spots from dataset
If you want to remove spots from a dataset (matrix in TSV) for instance to keep only spots inside the tissue. You can do so with the script adjust_matrix_coordinates.py
adjust_matrix_coordinates.py --counts-matrix stadata.tsv --outfile new_stdata.tsv --coordinates-file coordinates.txt
Where coordinates.txt will be a tab delimited file with 6 columns:
orig_x orig_y new_x new_y new_pixel_x new_pixel_y
Only spots whose coordinates in the file will be kept and then optionally you can update the coordinates in the matrix choosing for the new array or pixel coordinates.
The ST Pipeline generate useful statistical information in the LOG file but if you want to obtain more detail information about the quality of the data, you can run the following script:
st_qa.py --input-data stdata.tsv
If you want to perform quality stats on multiple samples you can run:
multi_qa.py --counts-table-files stdata1.tsv stadata2.tsv stdata3.tsv ...
Multi_qa.py generates violing plots, correlation plots/tables and more useful information and it allows to log the counts for the correlation.
You can see a more detailed documentation in the folder "doc_out".
You can see a real dataset obtained from the public data from the following publication (http://science.sciencemag.org/content/353/6294/78) in the folder called "data".
The ST pipeline is open source under the MIT license which means that you can use it, change it and re-distribute but you must always refer to our license (see LICENSE and AUTHORS).
If you use the ST Pipeline, please refer its publication:
ST Pipeline: An automated pipeline for spatial mapping of unique transcripts Oxford BioInformatics 10.1093/bioinformatics/btx211
For questions, bugs, feedback, etc.. you can contact Jose Fernandez Navarro firstname.lastname@example.org
The ST Pipeline depends on some Python packages that will be automatically installed during the installation process. You can see them in the file dependencies.txt
The ST Pipeline requires to have installed in the system the aligner STAR (minimum version 2.5.4 if you use a ST Pipeline version >= 1.6.0) : https://github.com/alexdobin/STAR
The ST Pipeline is recommended to be run on a computer with at least 32GB of RAM and 8 cpu cores.