[Brian Bushnell]

Oh - that's an intentional protection from overwriting files. Just delete the output file first or add the "overwrite" flag.

[lyw1]

high contaninants

Thanks.

Input is being processed as unpaired

Input: 385043 reads 10781204 bases.
Contaminants: 341911 reads (88.80%) 9573508 bases (88.80%)
Result: 43132 reads (11.20%) 1207696 bases (11.20%)

What is diffinition of contaminants? It looks very high.

[lyw1]

I need to read 30 nt for sequences. Miseq read 32 nt in sequencing. Thus many sequences have NN at last 2 positions. Does this relate to high contaminant rate?

[Brian Bushnell]

Are you using bbduk.sh? That's the only one that prints anything about contaminants. Can you show your specific command line?

Anyway, if you tried filtering out adapters and you got a result like that, it means you have almost no product and mostly adapter sequence.

[lyw1]

Yes, bbduk.sh.

Input is being processed as unpaired

Input: 385043 reads 10781204 bases.
Contaminants: 341911 reads (88.80%) 9573508 bases (88.80%)
Result: 43132 reads (11.20%) 1207696 bases (11.20%)

[Brian Bushnell]

Please give me the exact command line (what you typed before you hit enter).

[lyw1]

k=16 shows high contaminants than k=26

zheng@zheng-XPS-8500:~/Desktop/bbmap/20140916ngs$ bbduk.sh -Xmx1g in=probe48mix25fg_S7_L001_R2_001.fastq ref=ngs13template.fasta stats=probe48mix25fg_S7_L001_R2_001_26.txt k=26 fbm
java -ea -Xmx1g -cp /home/zheng/Desktop/bbmap/current/ jgi.BBDukF -Xmx1g in=probe48mix25fg_S7_L001_R2_001.fastq ref=ngs13template.fasta stats=probe48mix25fg_S7_L001_R2_001_26.txt k=26 fbm
Executing jgi.BBDukF [-Xmx1g, in=probe48mix25fg_S7_L001_R2_001.fastq, ref=ngs13template.fasta, stats=probe48mix25fg_S7_L001_R2_001_26.txt, k=26, fbm]

No output stream specified. To write to stdout, please specify 'out=stdout.fq' or similar.
Initial:
Memory: free=237m, used=14m

Added 13 kmers; time: 0.023 seconds.
Memory: free=228m, used=23m

Input is being processed as unpaired

Input: 159642 reads 4469976 bases.
Contaminants: 130724 reads (81.89%) 3660272 bases (81.89%)
Result: 28918 reads (18.11%) 809704 bases (18.11%)

Time: 0.197 seconds.
Reads Processed: 159k 811.47k reads/sec
Bases Processed: 4469k 22.72m bases/sec
zheng@zheng-XPS-8500:~/Desktop/bbmap/20140916ngs$ ^C
zheng@zheng-XPS-8500:~/Desktop/bbmap/20140916ngs$ bduk.sh -Xmx1g in=probe48mix25fg_S7_L001_R2_001.fastq ref=ngs13template.fasta stats=probe48mix25fg_S7_L001_R2_001_16.txt k=16 fbm
bduk.sh: command not found
zheng@zheng-XPS-8500:~/Desktop/bbmap/20140916ngs$ bbduk.sh -Xmx1g in=probe48mix25fg_S7_L001_R2_001.fastq ref=ngs13template.fasta stats=probe48mix25fg_S7_L001_R2_001_16.txt k=16 fbm
java -ea -Xmx1g -cp /home/zheng/Desktop/bbmap/current/ jgi.BBDukF -Xmx1g in=probe48mix25fg_S7_L001_R2_001.fastq ref=ngs13template.fasta stats=probe48mix25fg_S7_L001_R2_001_16.txt k=16 fbm
Executing jgi.BBDukF [-Xmx1g, in=probe48mix25fg_S7_L001_R2_001.fastq, ref=ngs13template.fasta, stats=probe48mix25fg_S7_L001_R2_001_16.txt, k=16, fbm]

No output stream specified. To write to stdout, please specify 'out=stdout.fq' or similar.
Initial:
Memory: free=237m, used=14m

Added 143 kmers; time: 0.028 seconds.
Memory: free=228m, used=23m

Input is being processed as unpaired

Input: 159642 reads 4469976 bases.
Contaminants: 151727 reads (95.04%) 4248356 bases (95.04%)
Result: 7915 reads (4.96%) 221620 bases (4.96%)

[Brian Bushnell]

So... that's telling you that you are getting matches between the stuff in your input file (probe48mix25fg_S7_L001_R2_001.fastq) and your reference file (ngs13template.fasta). And a shorter kmer will always find more matches in the presence of error.

probe48mix25fg_S7_L001_R2_001_26.txt will contain a list of which reference sequences were seen, and how many times they were seen.

[lyw1]

And a shorter kmer will always find more matches in the presence of error.

Here k=16 shows less match sequences than k=26

for k=16
Input: 159642 reads 4469976 bases.
Contaminants: 151727 reads (95.04%) 4248356 bases (95.04%)
Result: 7915 reads (4.96%) 221620 bases (4.96%)

for k=26
Input: 159642 reads 4469976 bases.
Contaminants: 130724 reads (81.89%) 3660272 bases (81.89%)
Result: 28918 reads (18.11%) 809704 bases (18.11%)

[Brian Bushnell]

In this case, the output is misleading... BBDuk assumes that the ref file is a file of contaminants because that's what I originally designed it for. So "Contaminants" actually means "Things that match the reference". I may change the wording eventually.

In other words, 95.04% of the reads matched the reference for K=16 and 81.89% did for K=26.

[lyw1]

Great, thanks.

Zheng

[lyw1]

Is there a size limitation for the referece sequences? It will not work when I add a 20 bp reference sequence.

[Brian Bushnell]

The size limit is the same as kmer length. So, if k=30, it will not work with anything less than a 30bp reference.

[lyw1]

Thanks.

How do you separate unambiguousReads and ambiguousReads in bbmap.sh?

[Brian Bushnell]

Ambiguously mapped reads get a "XT:A:R" tag in the sam output while unambiguously mapped get "XT:A:U".

You can also forbid ambiguously-mapping reads using the flag "ambig=toss", which will consider them unmapped.