[joa_ds]

Isn't there a feature to export unmapped reads to a file?

I always run bowtie and export unmapped and repeats using

--unfq unaligned.fastq --maxfa duplicates.fastq

taking a look at the size of both files compared to your original file gives you an approx idea of % unaligned/repeats

[Ben Langmead]

Hi tianell,

When I align certain short reads to reference using Bowtie, can I get a result message related to none-matched case??

I could not find an option to get a such result message.

I want to report even if certain short reads are not aligned to reference in order to use this information(not aligned!).

Sorry, no, there is no option to print such a message. I'll add this as a feature request. In the meantime, it's quite easy to deduce that number either by using the --un/--max options (and then counting), or by subtracting the reported number from the number of input reads.

Thanks,
Ben

[tianell]

Ben, help me..

Hi Ben,
I have a question for you about alignment result message.
When I align certain short reads to reference using Bowtie, can I get a result message related to none-matched case??

I could not find an option to get a such result message.

I want to report even if certain short reads are not aligned to reference in order to use this information(not aligned!).

I wil wait your answer, Ben. Thank you so much.

[Wind]

Thanks

Hi Ben,

Thanks for your advice. There were many 'N's in simulated data, so that they may interrupt paired-mapping. I'll try with other data sets. Thanks.

[Ben Langmead]

Bowtie is a nice tool for short read alignment I think. However, I found a problem in pair-end data mapping. I produced 75bp reads by simulating Illumina's high-throughput sequencing, and aligned them to the reference sequence. By the way, only few alignments, less than 10, are reported. As 1300000 alignments are reported with non paired-end mapping, probably it is wrongly mapped I think.
My option is "bowtie -p 8 -a -y -X 650 human -1 reads_1.fa -2 reads_2.fa output.map".

This is probably due to the -I/--minins, -X/--maxins, and/or --fr/--rf/--ff options being set incorrectly. Please double-check the manual's description of those options and verify that your invocation matches the way you've simulated your reads. Also, make sure the simulated read files are formatted correctly, with all mates lining up properly.

Thanks,
Ben

[Wind]

Bowtie is a nice tool for short read alignment I think. However, I found a problem in pair-end data mapping. I produced 75bp reads by simulating Illumina's high-throughput sequencing, and aligned them to the reference sequence. By the way, only few alignments, less than 10, are reported. As 1300000 alignments are reported with non paired-end mapping, probably it is wrongly mapped I think.
My option is "bowtie -p 8 -a -y -X 650 human -1 reads_1.fa -2 reads_2.fa output.map".

Can anybody tell me what is the problem?

[chuck]

Ben,

I used the latest version 0.10.1 and it still hangs. It seems to complete the job (or almost, I haven't verified that fact yet) and stops writing to the output file but then it never closes.

Do you want me to run the debug version or try the extra flags again?

Chuck

[chuck]

Hi Ben,

I've been teaching and not working on the data lately. I will give it a try soon.

I have a question for you about assembly quality evaluation, in two contexts.

1) to simply evaluate the quality of the assembly of the short reads against the reference sequences, beyond simple coverage
2) when there are actual differences between the sequenced genome and the reference genome, in finding indels and whatnot

I am looking AMOS, which seems to be one of the few that provide some kind of quality score for the assembly. Are you aware of others?

I am trying to quickly narrow my analysis down to those de novo contigs with good assembly scores. I proposed a simple metric in a manuscript and the reviewer suggested I use other 'standard' measures but gave no pointers as to which ones I should be using. Things are changing so fast it is hard to keep track of the 'standard'...

Thanks,
Chuck

[Ben Langmead]

And hence any suggestions to increase the mapping rate of Bowtie using -n options?

The main options used to adjust the sensitivity of mapping in Maq-like alignment mode are -n, -l, -e, --maxbts/-y. If there is a particular alignment you think Bowtie should be finding but isn't, please let me know and I can take a look.

Thanks,
Ben

[seq_GA]

Probably the -e limit again. See my previous post.

Ben

Hi Ben,

I am trying to get as many mapping as eland reports and trying to play around with Bowtie's parameters.
As you had suggested earlier, I tried using -e till 2000 to increase the mapping as good as eland but still Bowtie misses a lot of mappings when compared to eland.

-v option would give a comparable results ( I tested for read length 28 which is also the seed length) as eland but with the increasing number of Ns in the 3`end, it would be good to use -n option and try to allow any number of mismatches beyond seed length.

And hence any suggestions to increase the mapping rate of Bowtie using -n options?

Thanks.

[Ben Langmead]

Hi Chuck,

I tried bowtie remade with extraflags but it just did the same thing. Would there be a log file somewhere or something in the map file? I can't seem to find any additional output.

If you have a moment, could you try your run again using the latest version of Bowtie (0.10.1, released on Monday).

Thanks,
Ben

[frozenlyse]

Bowtie will index and align against references containing non-A/C/G/T characters, but alignments overlapping non-A/C/G/T characters in the reference are invalid and won't be reported.

Out of curiosity, what's the behavior you would like? E.g. if a C in a read were to align against a Y in the genome, would you like that to be considered a match, incurring no penalty against the alignment?

Thanks,
Ben

The reason out group would like this functionality is because we are investigating performing DNA methylation analysis via illumina bisulfite sequencing -> in this case C nucleotides in the normal genome will either be C or T nucleotides in the bisulfute converted genome.

So our preferred behavior would be to not penalise either the C or T (if the reference contained a Y at this position)

Anyway I find bowtie very useful, thanks for all your work!

[Ben Langmead]

Can you please explain me why there are more number of mapping when -v 2?

Probably the -e limit again. See my previous post.

Ben

[Ben Langmead]

with -v 3, Bowtie is also reporting one mapping location.

I want to use seedlength as 28(default) with 2 mismatches. hence I used -n 2 since I am comparing eland_28 and Bowtie results.

But still why Bowtie is not reporting?

Probably because the -e limit is disqualifying that alignment. If you'd like Bowtie to report alignments like that, try setting a higher -e than the default (70). -e is described in the Maq-like Policy section of the manual.

Ben

[seq_GA]

Hi Ben,
I did a quick comparison on with -v 2 and -n 2.

The reads are 35bps length and i used -3 6 to trim 3` sequences and hence my mappabale reads would be 28 in size in order for me to compare eland_28 results.

 bowtie -a -m 10 -v 2 --strata --best --solexa-quals  -p 15 -3 ../../Genome/hg18/hg18 ../s_1_sequence.txt out_aln.txt

When I look at the unque ly mapped tags with -v 2 is more than with -n 2.

Can you please explain me why there are more number of mapping when -v 2?

Thanks.