pXg: proteomics X genomics

Thank you, Dr. Jongmin Choi, for designing the logo!
Please note that the pXg-GUI version shared here is a temporary build. It is still under active development, and the final version will be released in the future <download>.

---

• About pXg
• Usage
  • Input
  • Output
    • pXg result
    • Unknown sequences
  • Amino acid variant table format
  • Command-line interface
    • List of parameters
    • Basic command
• Tutorial
  • RNA-Seq alignment
  • SAM/BAM preparation
  • Toy example
  • Run pXg
  • Run Percolator using the pXg results
  • IGV viewer
• TODO
  • GTF Export
• Citation

---

About pXg

pXg (proteomics X genomics) is a software tool that enables the reliable identification of both canonical and noncanonical MHC-I-associated peptides (MAPs) from de novo peptide sequencing by utilizing RNA-Seq data.

Usage

pXg can be integrated with any search engine, such as PEAKS and pNovo3. It was developed for the reliable identification of noncanonical MAPs from de novo peptide sequencing, but it can also be used to count the number of reads mapped to each peptide sequence.

Input

Input	Description	Format	Mandatory
Search result	A list of PSMs identified by a search engine (e.g. PEAKS, pNovo3, Casanovo)	TSV or CSV	Yes
Gene annotation	It must be the same file used in the read alignment (e.g. Gencode, Ensembl)	GTF	Yes
RNA-Seq reads	Mapped and unmapped RNA-Seq reads. The file must be sorted by coordinates. Multiple SAM/BAM files should be separated by comma (,)	SAM/BAM	Yes
Protein sequences	Canonical and contaminant protein sequences (e.g. UniProt)	Fasta	No

*pXg does not support pNovo3's flat-formatted output directly; you must convert it to CSV or TSV first.
*Since version 2.3.0, pXg supports multiple SAM/BAM files. The "Reads" column shows the sum of reads across all SAM/BAM files, and the per-file read counts are appended as additional columns at the end.

Output

Output	Description	Format	Mandatory
pXg result	This is a main output file and contains a list of identification as TSV format	TSV	Yes
pXg result for Percolator	This is a main output file and contains a list of identification as PIN format	PIN	Yes
Matched reads*	Matched reads to peptides passing all filters	SAM	No

*Although the pXg result contains PSM information with corresponding RNA-Seq read counts, it is not directly suitable for visualization.
The two additional output files (matched reads and peptides) can be used directly in IGV, making visualization easier.

pXg Result

Field	Description	Value
SpecID	Identifier of a spectrum	String
GenomicID	Identifier of genomic sequence	Integer
Label	Target (1) and decoy (-1) labels	1|-1
nDeltaScore	Normalized difference between main scores of current rank and top-rank peptides	Float
Rank	Rank of candidate peptides	Integer
GenomicLociCount	The number of genomic locations	Integer
AminoAcidVariant	An amino acid substitution in the format [position]:[original amino acid]>[altered amino acid]	String
InferredPeptide	Translated nucleotide sequence with a PTM annotation	String
InferredSequence	Translated nucleotide sequence without a PTM annotation	String
GenomicLoci	Genomic location of the peptide	String
Strand	Strand of matched sequence	+|-
ObservedLeftFlankNucleotide	Nucleotide sequence of the left flank of the peptide	String
ObservedNucleotide	Nucleotide sequence of the peptide	String
ObservedRightFlankNucleotide	Nucleotide sequence of the right flank of the peptide	String
ReferenceLeftFlankNucleotide	Reference nucleotide sequence of the left flank of the peptide	String
ReferenceNucleotide	Reference nucleotide sequence of the peptide	String
ReferenceRightFlankNucleotide	Reference nucleotide sequence of the right flank of the peptide	String
Mutations	Genomic information of mutations in the peptide	String
MutationStatus	Indication of alteration caused by the mutations	Altered|Same
TranscriptIDs	Matched transcript IDs	String
GeneIDs	Matched gene IDs	String
GeneIDCount	The number of matched gene IDs	Integer
GeneNames	Matched gene names	String
GeneNameCount	The number of matched gene names	Integer
PercentFullDistance	Proportion of start genomic loci in the longest transcripts (exons + introns)	Float
PercentExonDistance	Proportion of start genomic loci in the longest transcripts (exons)	Float
PercentCDSDistance	Proportion of start genomic loci in the longest transcripts (CDSs)	Float
FromCDSStartSite	Distance from the start site	String
FromCDSStopSite	Distance from the stop site	String
Events	Type of identified feature	String
EventCount	The number of events	Integer
FastaIDs	Matched identifiers in a given fasta sequences	String
FastaIDCount	The number of FastaIDs	Integer
Reads	Geometric mean of matched reads (or RPHM) across all SAM/BAM files	Float
MeanQScore	Mean of Phred scores	Float
IsReference	Reference (true) or non-reference (false) status	true|false
SAM/BAM file name	The number of matched reads (or RPHM) in each SAM/BAM file	Float

Unknown sequences

The unknown-sequence section contains sequence information from "unknown" events. Each header line begins with ">[PEPTIDE]", followed by the matched read information: the sequence identifier, genomic location (if available), full sequence, and matched sequence.

Amino acid variant table format

aaRNA	aaPeptide
W	F
W	M

Single amino acid variants (SAAVs) can arise after translation and therefore cannot always be detected at the RNA level.
pXg can account for SAAVs based on de novo peptide sequencing results.
As an example, we illustrate two SAAVs: W→F and W→M.

Command-line interface

List of Parameters

Option	Description	Value	Mandatory
gtf	GTF file path. We recommend using the same GTF file used for the alignment	String	Yes
bam	SAM/BAM file path. The file must be sorted by coordinate. Multiple SAM/BAM files should be separated by commas (,)	String	Yes
psm	PSM file path. The file is expected to be a proteomics search result from a de novo or database search engine	String	Yes
identifier_index	PSM identifier column indices (one-based). One or more indices can be specified, separated by commas. ex) 3,5,7	Integer	Yes
peptide_index	Peptide column index in the PSM file	Integer	Yes
charge_index	Charge state column index in the PSM file	Integer	Yes
output	Base output file name for pXg	String	Yes
comment	Specify the starting characters of comment lines to be ignored during processing. Lines beginning with these characters will be skipped. Default is #|@|%|MTD	String	No
min_score	Specify the minimum score threshold for peptide-spectrum matches (PSMs) to be included. Default is 0	float	No
count	Specify which reads to consider when counting read support. Default is primary	primary|all	No
sep	Specify the column separator. Possible values are csv or tsv. Default is tsv	tsv|csv	No
mode	Specify the library strandedness (default is auto). auto: auto-detected, paired-end only. fr: first-forward, second-reverse. rf: first-reverse, second-forward. r: reverse single-end. f: forward single-end. none: unstranded	f|r|fr|rf|auto|none	No
add_index	Specify the column indices of additional features to include in the PIN file. Multiple features can be added, separated by commas. ex) 5,6,7	Integer	No
il_equivalent	Controls whether pXg treats isoleucine (I) and leucine (L) as equivalent for peptide identification. Default is true	true|false	No
lengths	Peptide length range to consider (min-max, inclusive). Default is 8-15. ex) 8-13	Integer	No
fasta	Canonical sequence database used to conservatively assign noncanonical PSMs	String	No
rank	Number of ranked candidates to consider per scan. Default is 100 (effectively all ranked candidates)	Integer	No
aa_variant	File path of the amino acid variant table	String	No
output_sam	Report matched reads as a SAM file (true or false). Default is false	true|false	No
output_canonical	Report canonical peptides in the SAM output (true or false). Default is true	true|false	No
output_noncanonical	Report noncanonical peptides in the SAM output (true or false). Default is true	true|false	No
penalty_mutation	Penalty per mutation. Default is 1	Float	No
penalty_alternative_splicing	Penalty for alternative splicing. Default is 10	Float	No
penalty_5utr	Penalty for 5`-UTR. Default is 20	Float	No
penalty_3utr	Penalty for 3`-UTR. Default is 20	Float	No
penalty_ncrna	Penalty for noncoding RNA. Default is 20	Float	No
penalty_frameshift	Penalty for frameshift. Default is 20	Float	No
penalty_intron_retention	Penalty for intron retention. Default is 30	Float	No
penalty_intergenic_region	Penalty for an intergenic region. Default is 30	Float	No
penalty_asrna	Penalty for antisense RNA. Default is 30	Float	No
penalty_softclip	Penalty for softclipped reads. Default is 50	Float	No
penalty_unknown	Penalty for unmapped reads. Default is 100	Float	No
gtf_partition_size*	Size of the genomic region processed at once. Default is 5000000	Integer	No
sam_partition_size*	Number of reads processed at once. Default is 1000000	Integer	No
threads*	Number of worker threads. Default is 4	Integer	No

*These partition-size parameters affect memory usage and runtime. If your machine has limited memory, reduce these values.

Basic command

java -Xmx30G -jar pXg.jar \
--gtf [gene annotation file path] \
--bam [sorted SAM/BAM file path] \
--psm [de novo result file path] \
--fasta [protein sequence fasta file path] \
--identifier_index [index of file name column] \
--charge_index [index of charge state column] \
--peptide_index [index of peptide column] \
--score_index [index of search score column] \
--output [base output file name]

Tutorial

This tutorial walks through how to run pXg and estimate FDR from its results. It covers 1) running the STAR2 aligner with the 2-pass option, 2) preparing the SAM file from the alignment, 3) running pXg, and 4) post-processing steps, including running Percolator, merging the pXg result with the Percolator result, and estimating separate FDRs. Note that this tutorial does not cover how to run de novo peptide sequencing engines such as PEAKS, pNovo3, and Casanovo, nor how to generate deep-learning-based features.

RNA-Seq alignment

We recommend aligning FASTQ files with STAR2 using the two-pass alignment option from The Cancer Genome Atlas (TCGA) pipeline.

Sorted SAM/BAM preparation

Once you have the aligned BAM or SAM file, you MUST sort it by chromosomal coordinates.

Below is example code for preprocessing the SAM file using SAMtools:

samtools sort -o in.sorted.bam in.bam -@ 8
samtools index in.sorted.bam -@ 8

The resulting "in.sorted.bam" file is then used as input for pXg.

Toy example

This tutorial provides toy datasets — 1) de novo results, 2) in.sorted.sam, 3) gene annotation (GTF), and 4) a protein sequence FASTA file — in the tutorial folder, so you can try running the pXg pipeline yourself.

Run pXg

Using the toy datasets, you can run the pXg pipeline with the following command:

java -Xmx2G -jar pXg.v2.0.1.jar \
--gtf toy.gtf \
--sam toy.sorted.sam \
--psm toy.psm.csv \
--fasta toy.fasta \
--output toy \
--identifier_index 2,5 \
--peptide_index 4 \
--score_index 8 \
--charge_index 11 \
--add_index 15 \
--sep csv \
--mode none \
--threads 2

This should take about 2 minutes.

Note that the required memory (e.g., "-Xmx50G") depends on the size of the BAM file. In our experience, "-Xmx30G" is enough to handle a ~20GB file.

Run Percolator using the pXg results

Once you have the pXg result, you can add extra features — such as spectral similarity and delta retention time — as described in our manuscript. Even without these additional features, you can still run Percolator and estimate FDR from the pXg results.
We recommend using Percolator version >= v3.06.1, since earlier versions have an issue printing proteinIds.
Post processing codes are also provided in the tutorial folder (post_process.ipynb).

IGV viewer

When pXg finishes identifying peptides, the resulting GTF and SAM files are immediately available in the IGV viewer.

TODO

GTF Export

• Export the pXg result as a GTF file.

Citation

pXg: Comprehensive Identification of Noncanonical MHC-I–Associated Peptides From De Novo Peptide Sequencing Using RNA-Seq Reads. Seunghyuk Choi and Eunok Paek, Molecular & Cellular Proteomics 2024.