Skip to content

huishenlab/tranquilizer

Repository files navigation

Tranquilizer

Tranquilizer is a Deep Learning (DL) based tool to annotate, visualize the annotated reads, demultiplex for single-cell long-reads data including TRAQNUIL-seq and scNanoRNASeq and generate some inital QC plots.

I/O

The directory storing all the raw reads in fasta/fa/fasta.gz/fa.gz/fastq/fq/fastq.gz/fq.gz is provided as the input and the tool generates demultiplexed fasta files along with valid and invalid annotated .parquet files and some QC .pdf files as the outputs in two separate steps/commands (preprocessfasta and annotate-reads). Check out the examples drectory for both TRAQNUIL-seq and scNanoRNASeq datasets.

Usage

Preprocessing

To enhance the efficiency of the annotation process, Tranquilizer organizes raw reads into separate .parquet files, grouping them based on their lengths. This approach optimizes data compression within each bin, accelerates the annotation of the entire dataset, and facilitates the visualization of user-specified annotated reads without dependence on the annotation status of the complete dataset.

Example usage:

tranquilizer preprocessfasta /path/to/RAW_DATA/directory /path/to/OUTPUT/directory CPU_THREADS

It is recommended that you follow the directory structure as in the exmples.

Read length distribution

As an initial quality control metric, users may wish to visualize the read length distribution. The readlengthdist command facilitates this by generating a plot with log10-transformed read lengths on the x-axis and their corresponding frequencies on the y-axis. The output is provided in .png format in the /path/to/OUTPUT/directory/plots/ folder.

Example uage:

tranquilizer readlengthdist /path/to/OUTPUT/directory

Annotation, barcode correction and demultiplexing

Reads can be annotated, followed by barcode extraction, correction, and assignment to their respective cells (demultiplexing), using the single command annotate-reads. This command produces the following outputs:

  • Demultiplexed FASTA files: Located in /path/to/OUTPUT/directory/demuxed_fasta/.
  • Annotation metadata:
    1. Valid reads: /path/to/OUTPUT/directory/annotations_valid.parquet
    2. Invalid reads: /path/to/OUTPUT/directory/annotations_invalid.parquet
  • Quality control (QC) plots:
    1. barcode_plots.pdf
    2. demux_plots.pdf
    3. full_read_annots.pdf All QC plots are saved in /path/to/OUTPUT/directory/plots/.

Note: Before running the annotate-reads command, ensure you select the appropriate model for your dataset. If unsure, use the command tranquilizer availablemodels to view the available models.

Example usage:

tranquilizer annotate-reads MODEL_NAME /path/to/OUTPUT/directory /path/to/BARCODE_WHITELIST --chunk-size 100000 --portion full --njobs @CPU_threads

Read visualization

Annotated reads can be inspected independently of the annotate-reads process—either before or after successfully running the annotate-reads command—by providing their names to the visualize command. The resulting visualization is saved as a .pdf file in the /path/to/OUTPUT/directory/plots/ folder.

Example usage:

tranquilizer visualize MODEL_NAME /path/to/OUTPUT/directory --portion full --read-names READ_NAME_1,READ_NAME_2,READ_NAME3

** Installation instructions and pre-requisites coming soon **

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages