Spectral Consensus Spectrum Generation

spectral-consensus-spectrum-generation

Summary

Generate a consensus MS/MS spectrum by aggregating fragment m/z values and intensities across replicate spectra within a specified mass tolerance, producing a merged spectrum with fragment recurrence frequencies. This step is essential for denoising and downstream metabolite annotation in untargeted metabolomics.

When to use

After extracting and filtering top-TIC spectra for a given feature (e.g., using 80% TIC cutoff), when you need to identify stable, recurring fragments across multiple replicate MS/MS measurements and establish a baseline fragment frequency distribution for labeling and denoising individual spectra.

When NOT to use

• Input spectra have not been pre-filtered by top-TIC cutoff; consensus will reflect noise and low-quality spectra equally.
• Mass tolerance is not calibrated for your instrument type; grouping will either miss true co-eluting fragments or merge distinct ones incorrectly.
• Single replicate spectrum available for a feature; consensus spectrum requires multiple spectra to compute meaningful recurrence frequencies.

Inputs

• Top-TIC-filtered MS/MS spectra collection (Spectra object or equivalent) for a single feature
• Mass tolerance threshold (Da; typically 0.05)
• Folder path containing input and output data

Outputs

• Consensus spectrum dataframe with columns: mean m/z, mean intensity, fragment recurrence frequency
• Fragment count (total number of unique fragments after merging)
• Recurrence frequency vector for downstream labeling and filtering

How to apply

Input the top-TIC-filtered spectra collection for a single feature (e.g., 66 spectra after 80% TIC filtering from 83 original spectra). Apply the call_aggregate function with a mass tolerance parameter (default 0.05 Da) to group and merge fragment m/z values across all spectra, summing intensities for each grouped fragment and recording recurrence frequency (count of spectra in which each fragment appears). The output is a consensus spectrum dataframe containing mean m/z, mean intensity, and fragment recurrence count for each merged fragment. This frequency information becomes the reference standard for subsequent labeling of individual spectra and frequency-based denoising steps.

Related tools

• call_aggregate (Core function that merges fragments across top-TIC spectra within mass tolerance and computes recurrence frequencies) — https://github.com/BiosystemEngineeringLab-IITB/dures
• extract_raw_spectra (Prerequisite function that filters to top-TIC spectra and performs intra-spectrum grouping before consensus generation) — https://github.com/BiosystemEngineeringLab-IITB/dures
• Spectra (Bioconductor package providing MS/MS spectrum object representation and aggregation infrastructure)
• S4Vectors (Bioconductor package providing S4 class system used by Spectra and consensus spectrum data structures)
• dplyr (Data manipulation and aggregation within consensus spectrum dataframe construction)

Examples

l3 = call_aggregate(l2$sps_top_tic_2, 0.05, folder_path)

Evaluation signals

• Fragment count matches expected consensus spectrum size (e.g., 498 fragments for feature 1982 as reported in task_002).
• All fragments in the consensus spectrum have recurrence frequency ≥ 1 (each fragment appears in at least one replicate spectrum).
• Mean m/z and mean intensity values are within expected ranges for the instrument and ionization mode (e.g., positive mode).
• Recurrence frequency distribution is reasonable for the number of input spectra (e.g., max frequency ≤ number of top-TIC spectra used, typically 66 for 80% TIC cutoff on ~83 input spectra).
• Consensus spectrum has fewer or equal fragments than the largest individual input spectrum (merging should not create new fragments, only combine them).

Limitations

• Mass tolerance (0.05 Da) is empirically chosen and may not be optimal for all instruments; calibration via parameter tuning (Pareto front analysis) is recommended.
• Consensus generation assumes replicate spectra are comparable in quality and ionization efficiency; highly variable or outlier spectra can inflate fragment frequencies.
• Fragment recurrence frequency alone does not indicate biological significance; frequency-based denoising thresholds must be tuned relative to noise background.
• Consensus spectrum aggregates intensities across spectra, masking intensity variation that may correlate with sample composition or instrument drift.

Evidence

• [other] Generate consensus spectrum by applying call_aggregate to group fragments across all 66 top-TIC spectra with mass tolerance 0.05 Da to merge nearby m/z values and sum intensities.: "Generate consensus spectrum by applying call_aggregate to group fragments across all 66 top-TIC spectra with mass tolerance 0.05 Da to merge nearby m/z values and sum intensities."
• [other] Extract consensus spectrum dataframe containing mean m/z, mean intensity, and fragment recurrence frequencies for each of the 498 fragments.: "Extract consensus spectrum dataframe containing mean m/z, mean intensity, and fragment recurrence frequencies for each of the 498 fragments."
• [methods] In the third step, a consensus spectrum is generated using the top 80% TIC spectra, and the corresponding fragment frequencies are calculated: "In the third step, a consensus spectrum is generated using the top 80% TIC spectra, and the corresponding fragment frequencies are calculated"
• [readme] l3 = call_aggregate(l2$sps_top_tic_2, 0.05, folder_path): "l3 = call_aggregate(l2$sps_top_tic_2, 0.05, folder_path)"
• [methods] fragments in the reference spectra that fall within the specified m/z tolerance (tol) are grouped together: "fragments in the reference spectra that fall within the specified m/z tolerance are grouped together"