HybridFinder {RHybridFinder}

HybridFinder

Description

This function takes in three mandatory inputs: (1) all denovo candidates (2) database search results and (3) the corresponding proteome fasta file. The function's role is to extract high confidence de novo peptides and to search for their existence in the proteome, whether the entire peptide sequence or its pair fragments (in one or two proteins).

Usage

HybridFinder(
  denovo_candidates,
  db_search,
  proteome_db,
  customALCcutoff = NULL,
  with_parallel = TRUE,
  customCores = 6,
  export_files = FALSE,
  export_dir = NULL
)

Arguments

denovo_candidates

dataframe containing all denovo candidate peptides

db_search

dataframe containing the database search peptides

proteome_db

path to the proteome FASTA file

customALCcutoff

the default is calculated based on the median ALC of the assigned spectrum groups (spectrum groups that match in the database search results and in the denovo sequencing results) where also the peptide sequence matches, Default: NULL

with_parallel

for faster results, this function also utilizes parallel computing (please read more on parallel computing in order to be sure that your computer does support this), Default: TRUE

customCores

custom amount of cores strictly higher than 5, Default: 6

export_files

a boolean parameter for exporting the dataframes into files in the next parameter for the output directory, Default: FALSE, Default: FALSE

export_dir

the output directory for the results files if export_files=TRUE, Default: NULL, Default: NULL

Details

This function is based on the published algorithm by Faridi et al. (2018) for the identification and categorization of hybrid peptides. The function described here adopts a slightly modified version of the algorithm for computational efficiency. The function starts by extracting unassigned denovo spectra where the Average Local Confidence (assigned by PEAKS software), is equivalent to the ALC cutoff which is based on the median of the assigned spectra (between denovo and database search). The sequences of all peptides are searched against the reference proteome. If there is a hit then, then, the peptide sequence within a spectrum group considered as being linear and each spectrum group is is then filtered so as to keep the highest ALC-ranking spectra. Then, the rest of the spectra (spectra that did not contain any sequence that had an entire match in the proteome database) then undergo a "cutting" procedure where each sequence yields n-2 sequences (with n being the length of the peptide. That is if the peptide contains 9 amino acids i.e NTYASPRFK, then the sequence is cut into a combination of 7 sequences of 2 fragment pairs each i.e fragment 1: NTY and fragment 2: ASPRFK, etc).These are then searched in the proteome for hits of both peptide fragments within a same protein, spectra in which sequences have fragment pairs that match within a same protein, these are considerent to be potentially cis-spliced. Potentially cis-spliced spectrum groups are then filtered based on the highest ranking ALC. Spectrum groups not considered to be potentially cis-spliced are further checked for potential trans-splicing. The peptide sequences are cut again in the same fashion, however, this time peptide fragment pairs are searched for matches in two proteins. Peptide sequences whose fragment pairs match in 2 proteins are considerend to be potentially trans-spliced. The same filtering for the highest ranking ALC within each peptide spectrum group. The remaining spectra that were neither assigned as linear nor potentially spliced (neither cis- nor trans-) are then discarded. The result is a list of spectra along with their categorizations (Linear, potentially cis- and potentially trans-) Potentially cis- and trans-spliced peptides are then concatenated and then broken into several "fake" proteins and added to the bottom of the reference proteome. The point of this last step is to create a merged proteome (consisting of the reference proteome and the hybrid proteome) which would be used for a second database search. After the second database search the checknetmhcpan function or the step2_wo_netMHCpan function can be used in order to obtain the final list of potentially spliced peptides. Article: Faridi P, Li C, Ramarathinam SH, Vivian JP, Illing PT, Mifsud NA, Ayala R, Song J, Gearing LJ, Hertzog PJ, Ternette N, Rossjohn J, Croft NP, Purcell AW. A subset of HLA-I peptides are not genomically templated: Evidence for cis- and trans-spliced peptide ligands. Sci Immunol. 2018 Oct 12;3(28):eaar3947. <doi: 10.1126/sciimmunol.aar3947>. PMID: 30315122.

Value

The output is a list of 3 dataframes containing:

1. the HybridFinder output (dataframe) - the spectra that made it to the end with their respective columns (ALC, m/z, RT, Fraction, Scan) and a categorization column which denotes their potential splice type (-cis, -trans) or whether they are linear (the entire sequence was matched in proteins in the proteome database). Potential cis- & trans-spliced peptide are peptides whose fragments were matched with fragments within one protein, or two proteins, respectively.

2. character vector containing potentially hybrid peptides (cis- and trans-)

3. list containing the reference proteome and the "fake" proteins added at the end with a patterned naming convention (sp|denovo_HF_fake_protein) made up of the concatenated potential hybrid peptides.

See Also

read.fasta,s2c

Examples

## Not run: 
 hybridFinderResult_list <- HybridFinder(denovo_candidates, db_search,
 proteome, export = TRUE, output_dir)
 hybridFinderResult_list <- HybridFinder(denovo_candidates, db_search,
 proteome)
 hybridFinderResult_list <- HybridFinder(denovo_candidates, db_search,
 proteome, export = FALSE)

## End(Not run)