RDP classifier training issues

[greigite]

I am unsuccessfully attempting to retrain the RDP classifier, version 2.7 on the UNITE fungal database. I have tried reformatting and generating a custom tax file with entries like this, following the examples in the sample files folder:

0*Root*-1*0*rootrank
1*Fungi*0*1*domain
2*Chytridiomycota*1*2*phylum
3*Neocallimastigomycetes*2*3*class
4*Neocallimastigales*3*4*order
5*Neocallimastigaceae*4*5*family
6*Piromyces*5*6*genus
7*Piromyces_sp_I_GRL_10*6*7*species
8*Piromyces_sp_D_GRL_5*6*7*species
9*Piromyces_sp_AF_CTS_BTP1*6*7*species
10*Orpinomyces*5*6*genus
11*Orpinomyces_sp_NIANP60*10*7*species
12*Orpinomyces_sp_AF_CTS_BTO1*10*7*species
13*Orpinomyces_sp_AF_CTS_CHO3*10*7*species

I reformatted the headers in the UNITE file to look like this:

>Phaeoacremonium_pallidum|EU128053|SH114132.06FU|refs|r__Root;d__Fungi;p__Ascomycota;c__Sordariomycetes;o__Diaporthales;f__Togniniaceae;g__Phaeoacremonium;s__Phaeoacremonium_pallidum;

However, running the following command

java -jar rdp_classifier_2.7/dist/classifier.jar train --seq sh_
input.fasta -t rdp_tax_file.txt -o ~/rdp_classifier/

gives the following error:

Exception in thread "main" java.lang.NumberFormatException: For input string: ""
        at java.lang.NumberFormatException.forInputString(NumberFormatException.java:65)
        at java.lang.Integer.parseInt(Integer.java:504)
        at java.lang.Integer.<init>(Integer.java:677)
        at edu.msu.cme.rdp.classifier.train.TreeFactory.addSequencewithTaxid(TreeFactory.java:157)
        at edu.msu.cme.rdp.classifier.train.TreeFactory.addSequence(TreeFactory.java:141)
        at edu.msu.cme.rdp.classifier.train.ClassifierTraineeMaker.<init>(ClassifierTraineeMaker.java:72)
        at edu.msu.cme.rdp.classifier.train.ClassifierTraineeMaker.main(ClassifierTraineeMaker.java:171)
        at edu.msu.cme.rdp.classifier.cli.ClassifierMain.main(ClassifierMain.java:60)

Can anyone help me figure this out? Pretty sure I am missing something quite obvious, like for example changing the fasta headers to match the numbers in the tax file.

[yingeddi2008]

I am trying to use SILVA database to train RDP classifier to species level. I don't even know how the taxonomy file is generated. And in the RDP README file, it didn't say neither...I don't know how this retrain feature is useful

[bio_informatics]

Did you get any lead on this? I've similar query at this page.

[danova]

I found this script useful to retrain SILVA.

https://github.com/geraldinepascal/FROGS/blob/master/libexec/silva2retrainRDP.py

[yingeddi2008]

Still trying to figure out how to construct the taxonomy file

Hi everyone,

I am still trying to figure out how to construct the taxonomy file. What is the rule for putting the taxonomy ID and name together? Any leads will be appreciated.

Thanks,

Eddi