Open search returns non-tryptic peptides because Crystal-C adjusted a tryptic one to a semi-tryptic one.

[pisistrato]

I am trying to run a simple open search, using default settings. What I see is that a lot of the peptides returned in the table are not fully tryptic.

This is taken from the interact.pep.xml

<search_result>
<search_hit peptide="PIDFLEAK" massdiff="-0.00161" calc_neutral_pep_mass="931.50137" peptide_next_aa="G" num_missed_cleavages="0" num_tol_term="1" protein_descr="MYPROTEIN OS=MYO GN=MYP PE=1 SV=1" num_tot_proteins="1" tot_num_ions="16" hit_rank="1" num_matched_ions="10" protein="sp|X00XX0|MAYP1_MYO" peptide_prev_aa="N" is_rejected="0">
<search_score name="hyperscore" value="19.4"/>
<search_score name="nextscore" value="15.573"/>
<search_score name="expect" value="1.106497e-04"/>
<ptm_result localization="1" best_score_with_ptm="19.400997" score_without_ptm="13.815165" localization_peptide="pIDFLEAK" second_best_score_with_ptm="17.234322" ptm_mass="-114.04468"/>
<analysis_result analysis="peptideprophet">
<peptideprophet_result probability="0.9997" all_ntt_prob="(0.0000,0.9997,0.9980)">
<search_score_summary>
<parameter name="fval" value="7.8319"/>
<parameter name="ntt" value="1"/>
<parameter name="nmc" value="0"/>
<parameter name="massd" value="-0.002"/>
</search_score_summary>
</peptideprophet_result>
</analysis_result>
</search_hit>
</search_result>

I checked the database (including the reverse sequences), and peptide PIDFLEAK can only map to X00XX0, but the previous amino acid is N (--TIMERSSFEKNPIDFLEAK--).

Why this is happening? Am I missing something here?

[fcyu]

Could you share your log file, fasta file, mzML file, and the interact.pep.xml file with us?

Thanks,

Fengchao

[pisistrato]

Hi @fcyu You can get it from here https://syncandshare.lrz.de/getlink/fiSVzs4hxWfckRFRnyUACU/

[fcyu]

Somehow I couldn't download the files. Could you upload the files to https://www.dropbox.com/request/flObzxZqiLXxLWwFJVZk

Thanks,

Fengchao

[pisistrato]

Done, thanks!

[fcyu]

Could you also upload the fasta file?

[pisistrato]

I thought I did, sorry. It is now there.

[fcyu]

I think I might find out the reason. To confirm my hypothesis, could you upload your *.pepXML files (the ones without interact- prefix)?

[pisistrato]

Great. The file should be there now.

[fcyu]

Thanks for your prompt response, and the files.

I have confirmed that it was due to Crystal-C. Taking your PSM as an example, the following is from MSFragger (in Exp1_0268_R0095-01_S004226_G_B01_LFQ_02.pepXML)

<spectrum_query start_scan="16040" uncalibrated_precursor_neutral_mass="931.5009" assumed_charge="2" spectrum="Exp1_0268_R0095-01_S004226_G_B01_LFQ_02.16040.16040.2" spectrumNativeID="controllerType=0 controllerNumber=1 scan=16040" end_scan="16040" index="11979" precursor_neutral_mass="931.49976" retention_time_sec="1512.6200866699219">
<search_result>
<search_hit peptide="NPIDFLEAK" massdiff="-114.0445556640625" calc_neutral_pep_mass="1045.5443" peptide_next_aa="G" num_missed_cleavages="0" num_tol_term="2" protein_descr="CRISPR-associated endonuclease Cas9/Csn1 OS=Streptococcus pyogenes serotype M1 OX=301447 GN=cas9 PE=1 SV=1" num_tot_proteins="1" tot_num_ions="16" hit_rank="1" num_matched_ions="10" protein="sp|Q99ZW2|CAS9_STRP1" peptide_prev_aa="K" is_rejected="0">
<search_score name="hyperscore" value="19.4"/>
<search_score name="nextscore" value="15.573"/>
<search_score name="expect" value="1.106497e-04"/>
<ptm_result localization="1_2" best_score_with_ptm="19.400997" score_without_ptm="13.815165" localization_peptide="npIDFLEAK" second_best_score_with_ptm="17.234322" ptm_mass="-114.04468"/>
</search_hit>
</search_result>
</spectrum_query>

Then, Crystal-C changed the peptide to (in Exp1_0268_R0095-01_S004226_G_B01_LFQ_02_c.pepXML)

<spectrum_query start_scan="16040" uncalibrated_precursor_neutral_mass="931.5009" assumed_charge="2" spectrum="Exp1_0268_R0095-01_S004226_G_B01_LFQ_02.16040.16040.2" end_scan="16040" index="11979" precursor_neutral_mass="931.49976" retention_time_sec="1512.6200866699219">
<search_result>
<search_hit peptide="PIDFLEAK" massdiff="-0.00161" calc_neutral_pep_mass="931.50137" peptide_next_aa="G" num_missed_cleavages="0" num_tol_term="1" protein_descr="CRISPR-associated endonuclease Cas9/Csn1 OS=Streptococcus pyogenes serotype M1 OX=301447 GN=cas9 PE=1 SV=1" num_tot_proteins="1" tot_num_ions="16" hit_rank="1" num_matched_ions="10" protein="sp|Q99ZW2|CAS9_STRP1" peptide_prev_aa="N" is_rejected="0">
<search_score name="hyperscore" value="19.4"/>
<search_score name="nextscore" value="15.573"/>
<search_score name="expect" value="1.106497e-04"/>
<ptm_result localization="1" best_score_with_ptm="19.400997" score_without_ptm="13.815165" localization_peptide="pIDFLEAK" second_best_score_with_ptm="17.234322" ptm_mass="-114.04468"/>
</search_hit>
</search_result>
</spectrum_query>

That is why you saw PIDFLEAK in the interact.pep.xml file. The logic of changing NPIDFLEAK to PIDFLEAK is that 1) there are always a small fraction of semi-tryptic peptides in the sample; 2) there is a big delta mass which equals to the mass of N. After removing N, the mass diff became -0.00161.

Best,

Fengchao

[pisistrato]

I see, but why not reporting NPIDFLEAK with a modification (loss of N)? That would be better way of reporting the identification IMHO.

[fcyu]

That is because matching to PIDFLEAK (with ~0 delta mass and similar score) is more reasonable than NPIDFLEAK. The "loss of N" is likely not true.

Best,

Fengchao

[anesvi]

Semi-tryptic peptides (e.g. due to in-source fragmentation) are very common and not relevant to the main purpose of doing open PTM searches. See our Crystal-c manuscript. But if you do not want to reclassify such peptides, just uncheck Crystal-c