Needler

Code and algorithm to support our paper Needler: An Algorithm to Develop a Comprehensive Targeted MS Method Capable of Monitoring the Human Proteome.

Steps

Makefile should replicate the needed steps assuming a standard Linux environment (make, wget, etc) + Python Poetry are available. Alternatively, can use the provided .devcontainer which will create an environment with necessary tooling.

Makefile high-level outline:

• make build_env Configure the Python virtual environment with Poetry
• make download Download required datasets
  • Uniprot Swiss-Prot proteome fasta
  • A deep proteome and transcriptome abundance atlas of 29 healthy human tissues proteomics reference for detectable proteins per tissue.
• make munge pre-process the datasets
  • extract liver proteins from Tissue Atlas study
  • extract human proteins from Uniprot
  • digest proteins into tryptic peptides
  • filter tryptic peptides
    • sized 5-30 residues
    • distinct (represented by single protein)
    • does not contain uncommon amino acid (X or U)
    • does not contain methionine (commonly oxidized and is inappropriate for MS quantification)
    • each protein represented by >= 2 peptides (common MS quantitation criteria)
  • assign predicted iRT and retention time value to all retained peptide sequences
  • produce sub-proteomes for study: liver, kinase, dub
• make fit run the needler algorithm to produce targeted MS methods for each proteome. Recommend invoking with a -j <CPU_COUNT> option to run fits in parallel. This represents a significant amount of computation and is not advised to run on a single machine as it represents potentially years of dedicated computer time.

Data

Datasets contained within the repo itself:

• procal_supplementaltable_S1.csv. Supplementary data table from PROCAL: A Set of 40 Peptide Standards for Retention Time Indexing, Column Performance Monitoring, and Collision Energy Calibration of the 40 retention time standard peptide sequences chosen by the Kuster group.
• prosit_predict_irt.csv predicted indexed retention times (iRT) values for human peptide sequences using the Prosit: proteome-wide prediction of peptide tandem mass spectra by deep learning Proteome Tools calculator. Configuring the calculator takes a bit of effort, so committing computed values here.
• uniprot_pkinfam_202004.csv human kinases extracted from the Uniprot pkinfam resource

License

Code is available under an Apache 2 license.