This repository contains the code used in the manuscript Absent from DNA and protein: genomic characterization of nullomers and nullpeptides across functional categories and evolution by Georgakopolous-Soares et al. The folders contain scripts for identifying and analyzing nullomers from multiple genomes. Please note that this is not intended to be a package for nullomer analysis but rather a place where the code that was used for the manuscript can be accessed and reviewed.

This directory contains a set of Python scripts that enable the extraction of all nullomers from a search space.

• extract_kmers.py - this file scans the search space (FASTA format input) and counts the number of occurrences of every kmer. The range of kmer lengths to scan is provided as a variable. The results are saved in an output file. The example provided is for hg38 for kmer lengths between 1-15bps.
• possible_kmers.py - this script produces all possible kmers to search if they are nullomers or not.
• generate_all_counts.py - this script generates the final output file with the occurrences of each kmer including nullomers in the search space of interest. The output format is a tab delimited file with first column containing the kmer sequence and second column the occurrences in the search space.

This directory contains Julia files that are used to identify all possible nullomer creating single basepair deletion, insertion and substitution in the human genome.

• findAlmostNullomers.jl - this file scans the genome and for a given length it identifies all possible mutations creating kmers (does not have to be nullomers) that are created through a single basepair change. For the human genome it is recommended to use a high performance cluster for k>11. The flags -s and -e are meant to facilitate this by only considering a specific subset of the nullomers on the list.
• findAlmostNullpeptides.jl - similar to findAlmostNullomers.jl but operates on the proteome instead. Only reports substitutions. Again it is recommended to split up the process creating files that process 1,000 nullpeptides each using the -s and -e flags.
• findNullomerMutations.jl - this script takes as input a list of potential mutations (generated using findAlmostNullomers.jl) and a list of variants from gnomad.
• gnomad_raw_extract.sh - this script parses the (large) file from gnomad and extracts allele frequencies for all variants that involve a single bp substitution. This file is used by findNullomerMutations.jl. It will be useful to split the large file generated into smaller files, one per chromosome (using 23 for X and 24 for Y).
• gnomad_afr_extract.sh - this file extracts the variants that are population specific for afr. Minor modifications are required to extract variants specific for other populations.

Note that all of the above files will require editing to include suitable paths to genomes and other necessary resources.

The findAlmostNullpeptides.jl script identifies all amino acid substitutions that can result in the creation of a nullpeptide. However, due to constraints imposed by the genetic code, only a subset of these substitutions will be accessible through single basepair nucleotide substitions. To identify this subset of mutations, additional processing is required:

• Run annovar to identify the genomic regions that code for proteins perl /path/to/annovar/annotate_variation.pl -hgvs -buildver hg38 -dbtype wgEncodeGencodeBasicV33 /path/to/vcf/file/with/all/mutations/overlapping/coding/regions/all_cds_mutations.vcf /path/to/annovar/db/humandb/ -out /path/to/output/all_cds_mutations
• Extract the non-synonymous changes cut -f 2-8 /path/to/output/all_cds_mutations.exonic_variant_function | grep -e '^nonsyn' | cut -f 2-7 | awk {'printf ("%s\t%s\t%s\t%s\t%s\n", $2, $3, $1, $5, $6)'} > /path/to/output/all_cds_mutations_nonsyn.tsv
• Process the output from findAlmostNullpeptides.jl to put into a smaller file. The shorter file is assumed to contain three columns with the transcript name (in Gencode/Ensembl format), the mutation (in AAposAA format, eg Q432W) and the index of the nullpeptide being created from the is mutation (refers to the index of the list used for findAlmostNullpeptides.jl) grep ENST /path/to/file/nullpeptides5subspos.tsv | sed 's/\(ENST[[:digit:]]*\.[[:digit:]]*\)\t\([[:digit:]]*\)\t[A-Z]\([A-Z]\)[A-Z]:\([A-Z]\)\t[0-9]\t\([0-9]*\)/\1\t\3\2\4\t\5/' | sort -u > /path/to/file/nullpeptides5subspos_short.tsv
• To identify the mutations that correspond to these mutations run a small shell script. As this takes a long time it is recommended to split into smaller jobs on a high performance cluster and belowed it is assumed that the coordinates for the nullpeptide generating mutations are stored in files where 1,000 nullpeptides at a time were processed nfiles=$((n/1000 + 1)) ; for ((i=1; i<=$nfiles; i++)) ; do <command for submitting job to scheduler> run_find_all_nullpeptide_mutations.sh $i ; done Here it is assumed that $n contains the number of nullpeptides.
• The results can then be merged as cat /path/to/files/all_cds_nonsyn_nullpeptide5subs_* | sort -u > /path/to/files/all_cds_nonsyn_nullpeptide5subs_uniq.tsv
• The total number of mutations is found as wc -l /path/to/files/Gencode/all_cds_nonsyn_nullpeptide5subs_uniq.tsv
• We can add back the id of the nullpeptide as well grep -f /path/to/file/all_cds_nonsyn_one_transcript_nullpeptide5subs_uniq.tsv /path/to/file/nullpeptides5subspos_short.tsv > /path/to/file/all_cds_nonsyn_one_transcript_nullpeptide5subs_uniq_nullpepinds.tsv

This directory contains the Python scripts with which the occurrences of all kmers with 1bp Hamming distance from each nullomer were estimated. The input format is a tab delimited file with first column containing the kmer sequence and second column the occurrences in the search space.

• hamming_distance_nullomers_gen.py This script estimates the median number of occurrences of all 1bp Hamming distance kmers for each nullomer.

This directory contains the Python script with which simulations of sequences were performed controlling for mono / di / trinucleotide context.

• sim_sequences.py - this file generates the simulated FASTA files (n=100). We then extracted the nullomers and nullpeptides of the simulated genomes and proteomes using the nullomer_extraction and nullpeptide_extraction scripts.

This directory contains a set of Python scripts that enable the extraction of all nullpeptides from a proteome search space.

• extract_peptides.py - this file scans the proteome search space (FASTA format input) and counts the number of occurrences of every peptide kmer. The range of kmer lengths to scan is provided as a variable. The results are saved in an output file. The example provided is for the reference human proteome for nullpeptide kmer lengths between 1-7aa.
• possible_kmers.py - this script produces all possible peptide kmers to search if they are nullpeptides or not.
• generate_all_counts.py - this script generates the final output file with the occurrences of each peptide kmer including nullpeptides in the proteome search space of interest.