Aim 1: Evaluate the effect of ramp time on precursor identification

Aim 2: Access the contribution of OpenSWATH scores on pyprophet statistical analysis

Workflow

Workflow 1. Ran OpenSWATHWorkflow for 6 mzML files (ion mobility extraction window set to 0.08, 0.06 and 0.10)
2. Ran pyProphet statistical analysis (score, peptide, protein, and export)
* only 24/26 scores were considered for semi-supervised learning in pyProphet score
* ran pyProphet 6x for IM extraction window 0.08 and 3x for 0.06 and 0.10
3. Ran enerate_identification_report_ver2.py to get number of precursors, peptides and proteins from tsv file

Table: Precursor, Peptide and Protein Count

Increasing Ramp Time Improves Precursor Identification

table showing mean precursor counts and SD

RampTime Scheme mean_count sd
25 1 42250.6 3968.8513
25 2 1350.2 1686.0355
50 1 73829.6 234.9783
50 2 9428.6 4530.9002
100 1 77122.2 3658.9163
100 2 13301.6 5108.0177

Scheme1 captures about 50% of precursors found in Scheme2

Ramp Time (ms) Scheme Number of Replicates Number of precursors identified in all replicates
50 1 5 71327
50 2 4 6084
100 1 5 69408
100 2 5 6127

51.95595% of precursors found in 50ms Scheme 1 are also found in Scheme 2

56.426693% of precursors found in 100ms Scheme 1 are also found in Scheme 2

25ms was dropped for this analysis because of the high degree of variation between replicates (almost no overlaps between 25ms scheme2 replicates)
results from a replicate of 50ms was dropped because including it would significantly drop the number of precursors identified in all replicates

Majority of precursors identified in 25ms are also identified in 50ms and 100ms (Scheme1 only)

Ramp Time (ms) Number of Precursors identified in all 5 replicates
25 32960
50 71327
100 69408

96.8507282% of precursors in 25ms are found in 50ms
93.4557039% of precursors in 25ms are found in 100ms
50ms and 100ms shared 61458 of precurosrs (86.1637248% of 50ms and 88.5459889% of 100ms)

Table showing the mean weights of the 24 scores

Most scores weren’t heavily weighted in the semi-supervised learning

I filtered out scores between 0.5 and -0.5 (12 scores left) and ran the pyProphet analysis again for osw with IM extraction window 0.08, 0.06 and 0.10
only did the analysis for scheme 1 because still ran into null s value when running pyProphet score after score filter (same problem with merged files)
3 replicates #### 12 scores used for pyProphet analysis

Precursor count dropped after score filter

Expanding IM extraction window improves precursor identification for 25ms but not for 50ms and 100ms