OpenSwathDecoyGenerator remove duplicates #6054

jcharkow · 2022-04-20T21:41:56Z

Description

Addressing issue #5653

This commit adds a check to OpenSwathDecoyGenerator to ensure that a decoy sequence is not generated if it is the same as a target sequence or another decoy sequence.

This check is done with modifications meaning that the same peptide sequence with different modifications can still have unique decoys generated

Tests are added and tests are changed to reflect the new behaviour.

Checklist:

[ X ] Make sure that you are listed in the AUTHORS file
[ X ] Add relevant changes and new features to the CHANGELOG file
[ X ] I have commented my code, particularly in hard-to-understand areas
[ X ] New and existing unit tests pass locally with my changes
[ X ] Updated or added python bindings for changed or new classes. (Tick if no updates were necessary.)

How can I get additional information on failed tests during CI:

If your PR is failing you can check out

http://cdash.openms.de/index.php?project=OpenMS and look for your PR. If you click in the column that lists the failed tests you will get detailed error messages.
Or click on the action: e.g., for clang-format linting

Note:

Once you opened a PR try to minimize the number of pushes to it as every push will trigger CI (automated builds and test) and is rather heavy on our infrastructure (e.g., if several pushes per day are performed).

Advanced commands (admins / reviewer only):

/reformat (experimental) applies the clang-format style changes as additional commit
setting the label "NoJenkins" will skip tests for this PR on jenkins (saves resources e.g., on edits that do not affect tests)
commenting with rebuild jenkins will retrigger Jenkins-based CI builds

Previously if peptide did not end in K or R switchKR would replace the last AA with a random AA. Here, the behaviour is changed such that if K or R is not present, we keep the terminal AA. This new behaviour means that the same nontryptic peptide sequence will always generate the same decoy peptide.

Not sure how to add to documentation without changing the format of the method. So had to change return type from void to peptide

Add a check to check that decoy sequences generated do not happen to be target sequences (this will cause an error in OpenSwathWorkflow). Also added appropriate tests. Current method works, but is quite slow with larger libraries.

instead of searching using vector, search using unordered map for decoy sequences in targets which drastically speeds up runtime

this allows file to be converted to .tsv without error

This reverts commit 79a090f. Furthermore, also change switch KR back to a void function (revert 5321022 partially)

This reverts commit e7b5f5d. Change test back to the previous one because changed switchKR behaviour back to default.

replace tests output to reflect the new changes (to swtichKR for openswathdecoyGenerator) and for dropping decoy duplicates (MRMDecoy_test)

with switchKR feature can randomly get 2 peptide sequences that are not tryptic and almost identical generate the same decoy e.g. PEPTIDA --> DITPEPQ and PEPTIDE --> DITPEPQ (Q is randomly chosen for the final AA twice) - Add check to ensure that this decoy sequence is not added twice - modify openswathdecoy_test5 to test for this function - since hard to test the random nature of switchKR, add 2 duplicate target peptides and ensure the decoy is only generated once - Also add the same target peptide with a modification, and ensure that this peptide does get a decoy generated for it (because it is a different sequence)

src/openms/include/OpenMS/ANALYSIS/OPENSWATH/MRMDecoy.h

src/openms/source/ANALYSIS/OPENSWATH/MRMDecoy.cpp

timosachsenberg · 2022-04-26T15:02:19Z

src/openms/source/ANALYSIS/OPENSWATH/MRMDecoy.cpp

+              full_peptide_name += pep.sequence[loc];
+            }
+            // C-terminal and N-terminal modifications may be at positions -1 or pep.sequence
+            for (Size modloc = 0; modloc < pep.mods.size(); modloc++)


Having this inner loop is potentially very inefficient. How often is this function called?

This is called once per peptide sequence so it might be called a lot. However the inner loop should only be called maybe maximum of 3 times because theoretically there are not too many modifications on a peptide. This code block is actually taken from OpenMS tsv writer.

OpenMS/src/openms/source/ANALYSIS/OPENSWATH/OpenSwathTSVWriter.cpp

Lines 145 to 160 in c2555c3

String full_peptide_name = "";

for (int loc = -1; loc <= (int)pep.sequence.size(); loc++)

{

if (loc > -1 && loc < (int)pep.sequence.size())

{

full_peptide_name += pep.sequence[loc];

}

// C-terminal and N-terminal modifications may be at positions -1 or pep.sequence

for (Size modloc = 0; modloc < pep.modifications.size(); modloc++)

{

if (pep.modifications[modloc].location == loc)

{

full_peptide_name += "(UniMod:" + String(pep.modifications[modloc].unimod_id) + ")";

}

}

}

however this is probably not the most efficient way. Basically here I am just looking for a data structure which can hold modified peptide sequences to check for equality.

Running with a big library with over 2000000 it generates decoys in a reasonable time for me (e.g. 20 sec) so I think this implementation is good enough for now

src/openms/source/ANALYSIS/OPENSWATH/MRMDecoy.cpp

timosachsenberg · 2022-04-26T15:03:18Z

/reformat

timosachsenberg

some minor formatting / typos
potential inefficent generation of modified strings

Co-authored-by: Timo Sachsenberg <timo.sachsenberg@uni-tuebingen.de>

timosachsenberg · 2022-04-26T20:05:53Z

Btw you can add suggestions to a batch so that only one commit is created. Each commit starts a CI run of which we only have a limited amount per month.

jcharkow · 2022-04-26T20:27:54Z

Btw you can add suggestions to a batch so that only one commit is created. Each commit starts a CI run of which we only have a limited amount per month.

Sorry and thanks for letting me know. I will do this next time.

timosachsenberg · 2022-05-03T06:52:22Z

src/openms/source/ANALYSIS/OPENSWATH/MRMDecoy.cpp

+      {
+        if (pep.mods[modloc].location == loc)
+        {
+  	full_peptide_name += "(UniMod:" + String(pep.mods[modloc].unimod_id) + ")";


there are still some style errors. e.g. here no indentation. below it uses probably tabs instead of two spaces. maybe give it another quick check?

I will give try to give a quick fix to move this forward

can't push to Roestlab... I think you need to do the changes yourself. sorry

no problem I'll take a look

manually go through uncrustify changes and change spacing and tabbing accordingly.

jcharkow added 11 commits April 4, 2022 15:41

Add documentation on switchKR method

5321022

Not sure how to add to documentation without changing the format of the method. So had to change return type from void to peptide

Change test ouput for new switchKR method

e7b5f5d

[FEATURE] check generated decoys are not targets

7c03e0c

Add a check to check that decoy sequences generated do not happen to be target sequences (this will cause an error in OpenSwathWorkflow). Also added appropriate tests. Current method works, but is quite slow with larger libraries.

Speedup searching for decoys in targets

6673eb0

instead of searching using vector, search using unordered map for decoy sequences in targets which drastically speeds up runtime

[FIX] protein reference

6cd984e

this allows file to be converted to .tsv without error

Revert "change switchKR behaviour with nontryptic peptide"

40d55eb

This reverts commit 79a090f. Furthermore, also change switch KR back to a void function (revert 5321022 partially)

Revert "Change test ouput for new switchKR method"

017d991

This reverts commit e7b5f5d. Change test back to the previous one because changed switchKR behaviour back to default.

fix tests

4e1e8ad

replace tests output to reflect the new changes (to swtichKR for openswathdecoyGenerator) and for dropping decoy duplicates (MRMDecoy_test)

[MINOR] fix formatting

94eb7b8

jcharkow requested a review from hroest April 20, 2022 21:41

jcharkow changed the title ~~Addressing Issue https://github.com/OpenMS/OpenMS/issues/5653~~ OpenSwathDecoyGenerator remove duplicates Apr 20, 2022