problems with meta proteome analyzer

[bernt-matthias]

ping @chambm

Still having problems with MPA. Input is a 600MB fasta file and 441MB mgf. Might be a memory problem, but -Xmx 70GB is already quite a lot (actual available memory was 3 * 37GB).

Any ideas what we could try here?

stderr

Picked up _JAVA_OPTIONS: -Xmx70g -Xms256m
java.lang.NullPointerException
	at org.mapdb.DataOutput2.writeUTF(DataOutput2.java:148)
	at org.mapdb.serializer.SerializerString.serialize(SerializerString.java:15)
	at org.mapdb.serializer.SerializerString.serialize(SerializerString.java:11)
	at org.mapdb.serializer.SerializerArrayTuple.valueArraySerialize(SerializerArrayTuple.java:62)
	at org.mapdb.BTreeMapJava$NodeSerializer.serialize(BTreeMapJava.java:166)
	at org.mapdb.BTreeMapJava$NodeSerializer.serialize(BTreeMapJava.java:136)
	at org.mapdb.StoreDirectAbstract.serialize(StoreDirectAbstract.kt:243)
	at org.mapdb.StoreDirect.update(StoreDirect.kt:631)
	at org.mapdb.BTreeMap.put2(BTreeMap.kt:408)
	at org.mapdb.BTreeMap.put(BTreeMap.kt:292)
	at org.mapdb.BTreeMapJava$KeySet.add(BTreeMapJava.java:529)
	at de.mpa.io.fasta.index.OffHeapIndex.addToPeptideIndex(OffHeapIndex.java:130)
	at de.mpa.io.fasta.index.OffHeapIndex.generateIndex(OffHeapIndex.java:104)
	at de.mpa.io.fasta.index.OffHeapIndex.setupDatabase(OffHeapIndex.java:76)
	at de.mpa.io.fasta.index.OffHeapIndex.<init>(OffHeapIndex.java:58)
	at de.mpa.cli.CmdLineInterface.call(CmdLineInterface.java:222)
	at de.mpa.cli.CmdLineInterface.<init>(CmdLineInterface.java:79)
	at de.mpa.cli.CmdLineInterface.main(CmdLineInterface.java:335)

stdout

X!Tandem
DB: Merged and Filtered FASTA from data 104, data 103, and others sequences: 1260552
Spectrums:2020-06-10_LK_P24_HEW4-PA(API:9e655f38e58ae482) 
Tue Jan 26 11:12:23 CET 2021 Creating FASTA index file...
Tue Jan 26 11:31:23 CET 2021 FASTA index file creation finished...
Tue Jan 26 11:31:23 CET 2021 Creating FASTA decoy file...
Tue Jan 26 11:31:26 CET 2021 FASTA decoy file creation finished...
Tue Jan 26 11:31:26 CET 2021 Creating peptide index file...

I tried with reduced fasta files (down to 10K sequences) without success.

Turning off the option Peptide indexing (of FASTA database) circumvents the problem. But I do not know what this option is doing, i.e. if it influences the results. The publication tells that the indexing is described in the wiki .. but I can not find anything in the wiki.

Edit: This resulted in:

Picked up _JAVA_OPTIONS: -Xmx70g -Xms256m
java.lang.NullPointerException
	at de.mpa.task.instances.CreateDatabaseTask.run(CreateDatabaseTask.java:67)
	at de.mpa.task.TaskManager.run(TaskManager.java:66)
	at de.mpa.task.SearchTask.init(SearchTask.java:73)
	at de.mpa.task.SearchTask.<init>(SearchTask.java:44)
	at de.mpa.client.Client.runSearches(Client.java:134)
	at de.mpa.cli.CmdLineInterface.call(CmdLineInterface.java:251)
	at de.mpa.cli.CmdLineInterface.<init>(CmdLineInterface.java:79)
	at de.mpa.cli.CmdLineInterface.main(CmdLineInterface.java:335)

Edit: with Iterative searching and Peptide indexing (of FASTA database) disabled it seems to run through.

Also reported upstream compomics/meta-proteome-analyzer#32

[chambm]

Did you mean to ping someone else?

[bernt-matthias]

Ups. My bad :) Seems that I mixed up my memory ..