Collection of general scripts for "Varizella-Zoster Virus proteomic profiling" (Girault et al., 2025) study using in-hous packages

These in-house R and Julia packages are used by the VZV data analysis scripts:

• msglm, Bayesian Mixed Effects Linear Models for Mass Spectrometry Data (R):
  • v0.5.0
  • v0.6.0
• msimportr, importing and pre-processing of MaxQuant/Spectronaut output (R):
  • v0.3.0
• HierarchicalHotNet.jl, Julia implementation of Hierarchical HotNet method:
  • v1.1.0

Affinity-purification of V5-tagged VZV proteins in SK-N-BE2 cells (interactomes)

The analysis of affinity-purification data of V5-tagged VZV proteins in SK-N-BE2 cells requires msglm (v0.5.0) and msimportr (v0.3.0) packages and is performed in the following steps:

1. prepare_data_apms.R script
   • Reads the MaxQuant files ProteinGroups.txt and evidence.txt and prepares the data for statistical analysis.
   • Outputs:
     • msfull.RData containing the formated and annotated dataset, with protein groups and peptides intensities.
     • msglm.RData containing the formated and annotated dataset with protein groups intensities, per-MS run intensity normalisation factors, MS instrument noise model parameters, the GLM model description (matrices of effects, batch effects, and contrasts) including parameter priors.
2. msglm_fit_chunk_apms.R script is called for each protein group (job chunk) in the data set
   • Gets protgroup_id as an input
   • Extracts the data of a given a protein group from msglm.RData
   • Fits the Bayesian model defined in msglm.RData using msglm package
   • Outputs <job_chunk>.RData file with the msglm results
3. msglm_fit-chunks_apms.sge.sh is a shell script for the parallel execution of msglm_fit_chunk_apms.R on the compute cluster using SGE job scheduler.
4. assemble_fits_apms.R script assembles all individual <job_chunk>.RData files into a single report.
   • Extracts the significance of relevant model contrasts for each protein group.
   • The results are compiled into the Supplementary Table S3, Tab 1 - VZV-Host interactions and Tab 2 - VZV ORF baits.

Full proteome changes induced by the depletion of the MPP8 gene in SK-N-BE2 cells

The analysis of proteomic changes induced by the depletion of the MPP8 gene in SK-N-BE2 cells requires msglm package (v.0.6.0) and msimportr (v0.3.0).

1. prepare_data_MPP8_KO.R
   • Reads the MaxQuant files: peptides.txt and evidence.txt
   • Corrects protein groups using the protregroup.jl script to avoid splitting isoforms of the same gene into individual protein groups that differ only by a single specific peptide (see Material and Methods section for details).
   • Outputs:
     • msfull.RData containing the formated and annotated dataset with protein groups and peptides intensities.
     • msglm.RData containing the formated and annotated dataset with peptide intensities, per-MS run intensity normalisation factors, MS instrument noise model parameters, the GLM model description (matrices of effects) including parameter priors.
2. msglm_fit_chunk_MPP8_KO.R script is called for each protein group (job chunk) in the data set
   • Gets protregroup_id as an input
   • Extracts the data of a given a protein group from msglm.RData
   • Fits the Bayesian model defined in msglm.RData using msglm package
   • Outputs <job_chunk>.RData file with the msglm results
3. msglm_fit-chunks_MPP8_KO.lrz.sh is a shell script for the parallel execution of msglm_fit_chunk_MPP8_KO.R on the compute cluster using SLURM workload manager.
4. assemble_fits_MMP8_KO.R script assembles all individual <job_chunk>.RData files into a single report.
   • Extracts the significance of relevant model contrasts for each protein group.
   • These results are integrated in the compiled "Complementary omics datasets" Table S5

Integration of the interactome and effectome data by network diffusion analysis

The integration of VZV virus-host interactome (virus-host protein interaction) and effectome (proteomic changes induced by the expression of individual viral proteins) is done by diffusing the protein abundance perturbations over the global network of protein-protein and functional gene interactions within the host cell (ReactomeFI database is used). It is implemented in Julia and uses in-house HierarchicalHotNet.jl package that implements Hierarchical HotNet method and provides additional statistics for the network diffusion process.

1. hotnet_analysis.jl is the general script for the HotNet analysis:
   • Reads the interactome and effectome (the results of the msglm analysis).
   • Reads the ReactomeFI network of protein-protein and functional gene interactions.
   • Prepares the effectome-based nodes and edges weights for the HotNet network diffusion analysis of each viral bait (step 2).
   • Prepares the 1000 random permutations of node and edge weights per viral protein for step 3 and saves them in hotnet_perm_input.jlser.zst file.
   • Reads the network diffusion results from step 2 and the combined permutation statistics from step 4 (hotnet_perm_assembled_<viral_protein>.jlser.zst).
   • Identifies the significant interactions between the host proteins physically associated with viral proteins (interactome) and the proteins effected by viral proteins overexpression (effectome)
   • Outputs the Supplementary Table S4 - HotNet analysis results (FIXME) with the significant interactions between the host proteins and viral proteins.
2. hotnet_treestats_chunks.jl
   • Performs HotNet network diffusion to integrate interactome and effectome for each viral protein using unperturbed weights.
   • Calculates the Strongly Connected Component Tree statistics for each edge weight cutoff threshold.
   • Outputs the hotnet_treestats_<viral_protein>.jlser.zst file with the HotNet network diffusion results for each viral protein.
   • hotnet_treestats_chunk.lrz.sh is the associated SLURM job script for parallel execution on the compute cluster.
3. hotnet_perm_chunk.jl
   • Reads a block of random effectome weights permutations (job chunk) from hotnet_perm_input.jlser.zst generated at step 1.
   • Performs memory and computationally intensive network diffusion for each weights permutation of the given job chunk.
   • Calculates the Strongly Connected Component Tree statistics for each edge weight cutoff threshold.
   • Outputs the hotnet_perm_<job_chunk>.jlser.zst file.
   • hotnet_perm_chunk.lrz.sh is the associated SLURM shell script for the parallel execution on the compute cluster.
4. hotnet_perm_chunk_assemble.jl
   • assembles the HotNet permuted tree results (hotnet_perm_<job_chunk>.jlser.zst) generated at step 3
   • Outputs the combined permutation statistics (hotnet_perm_assembled_<viral_protein>.jlser.zst).