Pepars - Protein Engineering via PARallel Sequencing

Pepars is a Python package containing various utilities for dealing with parallel sequencing data (e.g. NGS FASTQ files) for protein engineering contexts.

Installation

You can install Pepars via pip:

pip install git+https://github.com/GradinaruLab/pepars.git

Functionality

Pepars is broken up into packages roughly based on functionality. The main packages are:

• alignment
• analysis
• fileio
• plotting
• utils
• simulation

Details about the functionality of each package are below.

Alignment

The alignment package contains functions for taking FASTQ files and extracting variant regions. The main element of the alignment package is the Aligner class, which is an abstract class that defines how alignment should function.

To perform alignment on a set of FASTQ files, instantiate one of the subclasses of Aligner (e.g. Perfect_Match_Aligner or Bowtie_Aligner), and then call the align function:

align(
    template,
    FASTQ_file_sets,
    alignment_parameters=None,
    progress_callback=None,
    reverse_complement_template=None):

See examples/alignment.ipynb for a typical use case.

The alignment parameters vary by aligner, as below:

Perfect_Match_Aligner Parameters

variant_sequence_quality_threshold=0,
mismatch_quality_threshold=0

Bowtie_Aligner Parameters

working_directory=os.getcwd(),
is_local=False,
output_frequency=1e5,
approach=None,
allow_insertions_deletions=False,
quality_threshold=0.0

Subclassing Aligner

To subclass the Aligner class, your class needs to implement the internal _align method.

Analysis

The analysis package has a variety of analyses relevant to parallel sequencing-based protein engineering experiments. Some examples are:

• amino_acids.get_amino_acid_codon_biases: Get the expected bias of amino acids for a degenerate nucleotide sequence.
• sequencing_reads.get_nucleotide_distribution: Get the distribution of nucleotides in a FASTQ file
• confidence.get_sequence_confidences: Get the normalized confidences of a list of sequences and their counts, based on a few different confidence metrics

Fileio

Some simple file wrappers, designed to make reading/writing CSV and sequence count files easier.

Plotting

A set of wrappers around Plotly, designed to make it take just one or a few lines of code to get nice plots, either interactively or exported. Most of the plotting functions in here internally call plotting.generate_plotly_plot, which is a catch-all Plotly wrapper that prints to screen, or writes to a file, or both. To have plots generate interactively, make sure start your notebook with plotting.init_notebook_mode()

Some useful plots:

• plotting.plot_histogram: A simple Plotly histogram wrapper
• plotting.plot_scatter: A simple Plotly scatter plot wrapper
• plotting.plot_count_distribution: Plot the count distribution of variant sequences over multiple samples
• DNA.plot_amino_acid_bias: Plot a heatmap of amino acid bias, given sequence counts and a template

Utils

A variety of potentially useful utility functions. Some highlights:

• DNA: All the typical bioinformatics stuff: IUPAC grammar, the genetic code, translation/complement functions
• FASTQ_File: An easy-to-use FASTQ file iterator. Operates seemlessly on both gzipped and raw FASTQ files, and lets you iterate by sequences, quality scores, or both
• FASTQ_File_Set: A convenient way to iterate line-by-line along multiple FASTQ files in parallel - e.g. for paired end reads
• Sequence_Trie: A fast storage and query data structure for sequence information. Takes advantage of a trie structure to rapidly search for/iterate over one-off or two-off mutants.
• AminoAcid: A class for each Amino acid - useful for extracting physical properties

Simulation

Currently, this is just some functions for generating random mutants and their counts based on a template. Useful for testing.

License

License information can be found in the LICENSE file