3D-CLUMP

3D-CLUMP (3D-CLustering by Mutation Position) is an unsupervised clustering of amino acid residue positions where variants occur, without any prior knowledge of their functional importance, in 3D space. This code is based on the original CLUMP.

This is a program developed collaboratively by the laboratories of Dr. Tychele N. Turner at Washington University in St. Louis http://turnerlab.wustl.edu/ and Dr. Rachel Karchin at Johns Hopkins University https://karchinlab.org/.

Current Maintainers: Jeffrey Ng, Dr. Turner Lab, jeffrey.ng@wustl.edu

Yilin Chen, Dr. Karchin Lab, yilinc5@stanford.edu

Please check out our preprint on medRxiv: https://www.medrxiv.org/content/10.1101/2024.02.02.24302238v1 and watch for our future publication.

3D-CLUMP Docker image can be found here:

docker pull jng2/testme:clump3d_slim_env

Or by building the env.yaml file found in the dockerfile folder.

USAGE:

python 3D-CLUMP.py -f inputfile -p protein_lengths -s structurepath

OPTIONS: -a allele_frequency(Default=1) Remove Mutations Greater than an allele frequency threshold. Default includes every variant.

-c inputfile_controls Input file for the controls. A set of controls is required to get a statistical significance with the clump score

-z number_of_permutations The number of permutations you want to perform for significance testing.

-m minimum_number_of_mutations(Default=5) The minimum number of mutations in a gene in order to perform CLUMP.

-n normalize(Default=No) Do you want to normalize based on protein length. Normalization was not used in the published results.

-t Output Column Titles

Permutation CLUMP works best when the input file only contains one gene.

The input file and the control input file are in the same format:

• Column 1: GENE_HUGO_ID Required
• Column 2: PROTEIN_ID Required: Must match Protein Id's provided in the protein length file
• Column 3: STUDY_NAME Used as a column placeholder in CLUMP scripts (Can use NA if unavailable)
• Column 4: AMINO_ACID_POSITION Required: Amino Acid position of the variant
• Column 5: CHROM Used as a column placeholder in CLUMP scripts (Can use NA if unavailable)
• Column 6: POSITION Used as a column placeholder in CLUMP scripts (Can use NA if unavailable)
• Column 7: REF Allele Used as a column placeholder in CLUMP scripts (Can use NA if unavailable)
• Column 8: ALT Allele Used as a column placeholder in CLUMP scripts (Can use NA if unavailable)
• Column 9: ALLELE_FREQUENCY Required column. Need to add a value between 0 and 1. If you do not know it can just be 0 unless you are actually using the allele frequency feature of CLUMP.
• Column 10:DOMAIN Optional column (can be NA)

The protein length file is in the format: PROTEIN_ID LENGTH

You can find an updated protein length file: protein.2.length.2023.txt

Example runs

When running, please make ensure that /opt/conda/envs/snakemake/bin is in the PATH. It is set by default, but if your cluster overwrites the Docker image environment pathways, this will need to be either added to the path or by activating the snakemake conda environment.

Python docker run:

docker run  -v "/path/to/github_repo:/runme" -v "/path/to/pdb/files:/proteins" jng2/testme:clump3d_slim_env  python3 /runme/3D-CLUMP.py -s /proteins/ -f /runme/mutation_input -p /runme/protein.2.length.2023.txt -c /runme/control_input -m 5 -z 1000 

Output:

SMARCA2	NP_001276325	ndd_out	2.7132983495149348	1.0	0.0	0.8383716972664214	3.5516700467813562

Running high throughput CLUMP using snakemake:

Snakemake CLUMP

Local

snakemake -s case.control.snake --cores 2

Submit to cluster

snakemake --cluster 'your cluster information' -j 100 -w 30 -k -s case.control.snake