How can I format RDP database to be used in a BLAST search?

[miguelangel]

Hello there and thank you for your welcome!

I have to format RDP 16S bacterial database in fasta format (downladed from here:http://rdp.cme.msu.edu/misc/resources.jsp) to fit in a BLAST search, carried out with QIIME command 'assign_taxonomy.py'.

I need to create an index from the fasta file and I have read that this can be done by using 'formatdb' in the BLAST standsalone program, but when I try to do it I always get a message like this one:

[formatdb 2.2.22] ERROR: RDP_11_2_index.txt.nhrOutput
Blast-def-line-set.E.<title>
Invalid value(s) [9] in VisibleString [uncultured bacterium; DolOr_72351#Lineage=Root;rootrank;Bacteria;domain;unclassified_Bacteria; ...]

However, I get a .nhr file, but with data in this shape:

S000655540¢Ä0Ä0ĆÄňuncultured actinobacterium; GASP-KA1W3_B01#Lineage=Root;rootrank;Bacteria;domain;"Actinobacteria";phylum;Actinobacteria;class;Acidimicrobidae;subclass;Acidimicrobiales;order;"Acidimicrobineae";suborder;Acidimicrobiaceae;family;Ilumatobacter;genus°Ä0ĆİÄ


With unknown characters that does not allow to use it with BLAST option or in any other BLAST search.

Could any one help me with this issue? I am really stuck at this step...


Thanks a lot

MA

[miguelangel]

I have tried to do the same but using 'makeblastdb' and now I get this different error:

Error: (803.7) Blast-def-line-set.E.title
Bad char [0x9] in string at byte 38
uncultured bacterium; L2Sp-13 Lineage=Root;rootrank;Bacteria;domain;"Actinobacteria";phylum;Actinobacteria;class;Acidimicrobidae;subclass;Acidimicrobiales;order;"Acidimicrobineae";suborder;Acidimicrobiaceae;family;Ilumatobacter;genus

And a .nhr almost equal to the one generated with 'formatdb'.

I am sure that the problem is in the format of the original fasta file, that looks like this entry:

>S000655540 uncultured bacterium; L2Sp-13 Lineage=Root;rootrank;Bacteria;domain;"Actinobacteria";phylum;Actinobacteria;class;Acidimicrobidae;subclass;Acidimicrobiales;order;"Acidimicrobineae";suborder;Acidimicrobiaceae;family;Ilumatobacter;genus
ggaatcttgcgcaatgggcgaaagcctgacgcagcaacgccgcgtgcgggatgaaggccttcgggctgtaaaccgctttc
agcaggaacgaaaatgacggtacctgcagaagaaggagcggccaactacgtgccagcagccgcggtgacacgtaggctcc
aagcgttgtccggatttattgggcgtaaagagctcgtaggcggttgagtaagtcgggtgtgaaaactctgggcttaaccc
ggagacgccatccgatactgctctgactagagttcaggaggggagtggggaattcctagtgtagcggtgaaatgcgcaga
tattaggaggaacaccggtggcgaaggcgccactctggactgaaactgacgctgaggagcgaaagcatgggtatcaaaca
ggattagataccctggtactccatgccgtaaacggtgggcactaggtgtgggttccaactaacgggatccgcgccgtcgc
taacgcattaagtgccccgcctggggagtacggtcgcaagactaaaactcaaatgaattgacgg


Any idea of how can I change this format to fit into the formatdb/makeblastdb commands?

Thanks again

[GenoMax]

Problem is likely a tab character (between the S* and the rest of the header?) (based on the 0x9 code in your error). Also ID header line is probably wrapping on to second line (unless your copy/paste did that). You will likely need to reformat the headers.

[GenoMax]

Using the "release11_2_Bacteria_unaligned.fa" file downloaded from the link you posted I was able to create the indexes using makeblastdb (v. 2.2.29+) without the errors you saw. I did

$ makeblastdb -dbtype nucl -in release11_2_Bacteria_unaligned.fa

I got a certain number of errors (below), which may or may not indicate a real problem http://www.acgt.me/blog/2014/5/15/fu...rom-ncbi-blast

Error: (1431.1) FASTA-Reader: Warning: FASTA-Reader: First data line in seq is about 45% ambiguous nucleotides (shouldn't be over 40%)

[miguelangel]

Thanks a lot

I got exactly the same errors, so I will try if with these new files are properly formated to run BLAST.

[GenoMax]

I tried a test blast with a few sequences from the RDP fasta file. Worked without any problems.

If you do not need all the extra stuff in the fasta header ID you could remove most of it using the following command (leaving the S* ID's)

$ sed -e 's/>* .*$//' release11_2_Bacteria_unaligned.fa > release11_2_Bacteria_unaligned_truncated_header.fa

Then build the indexes from the new file.