arraylib-solve 0.10.0

Tool to computationally deconvolve combinatorially pooled arrayed random mutagenesis libraries

arraylib-solve

Introduction

arraylib-solve is a tool to deconvolve combinatorially pooled arrayed random mutagenesis libraries (e.g. by transposon mutagenesis). In a typical experiment generating arrayed mutagenesis libraries, first a pooled version of the library is created and arrayed on a grid of well plates. To infer the identities of each mutant on the well plate, wells are pooled in combinatorial manner such that each mutant appears in a unique combination of pools. The pools are then sequenced using NGS and sequenced reads are stored in individual fastq files per pool. arraylib-solve deconvolves the pools and returns summaries stating the identity and location of each mutant on the original well grid. The package is based on the approach described in [1].

Installation

To install arraylib-solve first create Python 3.8 environment e.g. by

conda create --name arraylib-env python=3.8
conda activate arraylib-env

and install the package using

pip install arraylib-solve

arraylib-solve uses bowtie2 [2] to align reads to the reference genome. Please ensure that bowtie2 is installed in your environment by running:

conda install -c bioconda bowtie2

How to run arraylib-solve

To run arraylib-solve on a library deconvolution experiment with default parameters run:

arraylib-run <input_directory> <experimental_design.csv> -c <number_of_cpu_cores_to_use> -gb <path_to_genbank_reference> -br <path_to_bowtie2_indices> -t <transposon_sequence> -bu <upstream_sequence_of_barcodes> -bd <downstream_sequence_of_barcodes>

Input parameters

Required parameters:

• input_dir: path to directory holding the input fastq files
• exp_design: path to csv file indicating experimental design (values should be separated by a comma). The experimental design file should have columns, Filename, Poolname and Pooldimension. (see example in tests/test_data/full_exp_design.csv)
  • Filename should contain all the unqiue input fastq filenames.
  • Poolname should indicate to which pool a given file belongs. Multiple files per poolname are allowed.
  • Pooldimension indicates the pooling dimension a pool belongs to. All pools sharing the same pooling dimension should have the same string in the Pooldimension column.

An example of how an exp_design file could look like:

Filename	Poolname	Pooldimension
column1.fastq	column1	columns
column2.fastq	column2	columns
row1.fastq	row1	rows
row2.fastq	row2	rows
platerow1.fastq	platerow1	platerows
platerow2.fastq	platerow2	platerows
platecol1.fastq	platecol1	platecols
platecol2.fastq	platecol2	platecols

• -gb path to genbank reference file
• -br path to bowtie index files, ending with the basename of your index (if the basename of your index is UTI89 and you store your bowtie2 references in bowtie_ref it should be bowtie_ref/UTI89). Please visit https://bowtie-bio.sourceforge.net/bowtie2/manual.shtml#the-bowtie2-build-indexer for a manual how to create bowtie2 indices.
• -t transposon sequence (e.g. AGATGTGTATAAGAGACAG)
• -bu upstream sequence of barcode (e.g. CGAGGTCTCT)
• -bd downstream sequence of barcode (e.g. CGTACGCTGC)

Optional parameters:

• -mq minimum bowtie2 alignment quality score for each base to include read
• -sq minimum phred score for each base to include read
• -tm number of transposon mismatches allowed
• -thr threshold for local filter (e.g. a threshold of 0.05 would filter out all reads < 0.05 of the maximum read count for a given mutant)

Output

arraylib-solve outputs 4 files:

• count_matrix.csv: Read counts per pool for each mutant.
• filtered_matrix.csv: Read counts per pool for each mutant, but mutants with barcodes with low read counts for a given genomic location are filtered out.
• mutant_location_summary.csv: A summary of mutants found in the well plate grid, where each row corresponds to a different mutant.
• well_location_summary.csv: A summary of the deconvolved well plate grid, where each row corresponds to a different well.

References

[1] Baym, M., Shaket, L., Anzai, I.A., Adesina, O. and Barstow, B., 2016. Rapid construction of a whole-genome transposon insertion collection for Shewanella oneidensis by Knockout Sudoku. Nature communications, 7(1), p.13270.
[2] Langmead, B. and Salzberg, S.L., 2012. Fast gapped-read alignment with Bowtie 2. Nature methods, 9(4), pp.357-359.