PySmooth

Here, we present a python implementation of SMOOTH (van Os,H. et al. (2005) ) called PySmooth which offers an easy-to-use command line interface and solves the drawbacks of SMOOTH. PySmooth reads the input genotype file and identifies singletons based on the algorithm described in SMOOTH with some modifications to allow four genotype codes, and flexible parameters. Unlike SMOOTH which doesn’t correct the singletons and missing data, PySmooth corrects genotype errors using a k-nearest algorithm. At each step, PySmooth generates summary files and visualizations that can be inspected by the user for further interpretation.

Installation and Dependencies

PySmooth has been tested with Python 3.8.12 version. It should work with Python >= 3.0 version. We recommend installing the anaconda python distributon. Download anaconda python distribution from https://www.anaconda.com/products/distribution and install following the instructions provided.

PySmooth depends on the following python libraries. These libraries are already included in the anaconda distribution. Therefore, you do not need to install them.

• numpy
• Pandas
• Sklearn
• matplotlib

You can simply download the following scripts from PySmooth GitHub page and put them in a single folder.

• utilities.py
• smooth.py
• ImputeMissingGenotype.py
• run_smooth.py

PySmooth can be executed by running the script run_smooth.py

Running run_smooth.py

Input Genotype File format

The First row is the header. Each row represents a unique marker.

The genotype file MUST have the following columns:

• Column 1: Chromosome name.
• Column 2: Genomic Position of the marker in the chromosome. For each chromosome,column 2 MUST already be sorted in ascending order.
• Column 3: Identification id of the marker location.
• Column 4: Reference allele in the reference genome if known or can be left blank cell.
• Column 5: Alternate allele if known or blank cell.
• Column 6 and beyond: Genotype code for the individuals in the marker location. Four codes can be used. A: Reference parent homozygous, B: Alternatte parent homozygous, H: heterozygous, U: missing data.

A screeshot of a portion of an example input file is shown below

Running PySmooth

PySmooth takes the following arguments

• -i or --input: Name of the input genotype file. This MUST be provided.
• '-o' or '--output': Prefix to name of output files to be generated. If not provided, default is test.
• -c or --chr: list of chromosome names to perform analysis on. Names should be separated by comma (e.g chr1,chr2,chr3). Default is to run through all the chromosomes in the genotype file.
• -l or --lower: Lowest threshold for identifying singletons. Default is 0.70.
• -uor --upper: Highest threshold for identifying singletons. Default is 0.98.
• -g or --gap: PySmooth iteratively identifies singletons starting with the highest threshold till the lowest threshold. This parameter is used to decreased the threshold at each iteration. Default is 0.02.
• -k : number of nearest neighbors to be used to assign correct genotype to singleton or missing item. Default value is 30.

First, change working directory to the folder where the PySmooth scripts are stored. You can do that by simply typing the following command in the terminal, or command prompt, or anaconda command prompt depending on your python installation or OS.

cd <path to where PySmooth scripts are stored>

Once the working directory is set, shown below are two examples of running PySmooth.

python run_smooth.py -i <path to the genotype file>/my_genotype_file.csv

• In the example folder, there is an example genotype input file called my_genotype_file.csv

The code above will analyze each chromosome detected and generate all output files with prefix test in the folder <path to the genotype file>

python run_smooth.py -i <path to the genotype file>/my_genotype_file.csv -o <path to output folder>/my_output -c chr1 -l 0.80 -u 0.98 -g 0.02

The code above will analyze for chromosome chr1and generate all output files with prefix my_output in the folder <path to output folder>.

Outputs

For each chromosome, PySmooth Generates the following outputs.

• Three summary csv files: <output>_<chr>.stats.csv, <output>_<chr>_singletons_stats.csv, and <output>_<chr>_imputed_stats.csv that contain % of homozygous, heterozygous calls for each individual for the raw genoytpe file, after singleton detection, and after error correction. An example file for the <output>_<chr>_singletons_stats.csv is shown below. The files not only indicate the number of singletons detected in each sample but also the fraction from each category of genotype call detected as singletons.

  

• Three bar plot png files: <output>_<chr>.stats.png, <output>_<chr>_singletons_stats.png, and <output>_<chr>_imputed_stats.png bar plot files that contains % of homozygous, heterozygous calls for each individual for the raw genoytpe file, after singleton detection, and after error correction, respectively. Example images are shown below.

• Three heatmap files: <output>_<chr>.heatmap.png, <output>_<chr>_singletons_heatmap.png, and <output>_<chr>_imputed_heatmap.png that visualize a color-coded image of different genotype codes in the original file, after singleton detection, and after error correction, respectively. Example images are shown below.

• <output>_<chr>_singletons.csv: genotype file with singleton detected. Singletons are marked as S.
• <output>_<chr>_imputed.csv: genotype file after error correction.

References

van Os,H. et al. (2005) SMOOTH: a statistical method for successful removal of genotyping errors from high-density genetic linkage data. Theor. Appl. Genet., 112, 187–94.