Aligning Illumina paired-end data with read length 250 bp

[opulcy97]

Hi All,

I am quite new in this forum and newbie bioinformatician. I have a very basic question to ask. I have illumina paired-end genomic data with read length 250 bp. I tried some alignment tools to align and map the reads with the reference genome.

At first I used Novoalign. But I found out that it supports maximum read length 150 bp. Though I posted this problem to Novoalign's forum and they advised me that -n 250 would work. It didn't work actually.

Then I was exploring BWA. However, BWA only supports maximum read length 200 bp. On the other hand BWA-SW supports more read length, which I think maximum 1000 kbp or something. But BWA-SW doesn't support pairea-end data.

Then I tried Stampy and it seems taking ages to align and map the data. It took 30 hours and eventually got killed by the system. I think it would not be a wise idea to use a too that takes more than 30 hours to align and map.

I am now at my wits end. Can anyone please tell me, which aligner I can use to align and map paired-end data with read length 250 bp?

Many thanks in advance.

Opulcy

[lh3]

BWA-SW supports paired-end reads. The online page needs to be updated. I would say a better choice is BWA-MEM available from the bwa gitbub, which will be release in the next couple of days (or even today if I can finish the final testing). The released bwa-0.7.0 has a bug that may cause segmentation fault in rare cases.

Bowtie2 is also a good choice, though as the developer (so could be biased), I think bwa-mem is more accurate. Here is the ROC curve (PDF) for those who are familiar with the metric.

EDIT: bwa-0.7.1 has been released. It fixes a few bugs in 0.7.0.

[mastal]

Aligning Illumina paired-end data with read length 250

Have you tried Bowtie2?

[opulcy97]

Thanks guys. Your help is much appreciated. I will try BWA-MEM and let you know if anything comes up.

Regards,
Opulcy

[opulcy97]

@Ih3

I ran BWA-MEM and it worked wonderfully as expected. Thanks a lot for your help.

Regards,
Opulcy

[lorendarith]

BWA-SW supports paired-end reads. The online page needs to be updated. I would say a better choice is BWA-MEM available from the bwa gitbub, which will be release in the next couple of days (or even today if I can finish the final testing). The released bwa-0.7.0 has a bug that may cause segmentation fault in rare cases.

Bowtie2 is also a good choice, though as the developer (so could be biased), I think bwa-mem is more accurate. Here is the ROC curve (PDF) for those who are familiar with the metric.

EDIT: bwa-0.7.1 has been released. It fixes a few bugs in 0.7.0.

Not being aware of this, um.. I aligned 250bp reads with BWA 0.6.2 and I did not spot any errors or anything out of the usual when aligning 100bp reads. Are my alignments then valid at all? I got an insert size which corresponds to the library prep and all...

:/

[lh3]

Not being aware of this, um.. I aligned 250bp reads with BWA 0.6.2 and I did not spot any errors or anything out of the usual when aligning 100bp reads. Are my alignments then valid at all? I got an insert size which corresponds to the library prep and all...

:/

It is not about being valid or not. You can use bwa-backtrack, the first algorithm, for 250bp reads, but bwa-mem will be better.

[shi]

Hi All,

I am quite new in this forum and newbie bioinformatician. I have a very basic question to ask. I have illumina paired-end genomic data with read length 250 bp. I tried some alignment tools to align and map the reads with the reference genome.

At first I used Novoalign. But I found out that it supports maximum read length 150 bp. Though I posted this problem to Novoalign's forum and they advised me that -n 250 would work. It didn't work actually.

Then I was exploring BWA. However, BWA only supports maximum read length 200 bp. On the other hand BWA-SW supports more read length, which I think maximum 1000 kbp or something. But BWA-SW doesn't support pairea-end data.

Then I tried Stampy and it seems taking ages to align and map the data. It took 30 hours and eventually got killed by the system. I think it would not be a wise idea to use a too that takes more than 30 hours to align and map.

I am now at my wits end. Can anyone please tell me, which aligner I can use to align and map paired-end data with read length 250 bp?

Many thanks in advance.

Opulcy

Dear Opulcy,

The Subread aligner can align reads of up to 1200bp long, and it can map paired-end reads. It's performance in mapping 200bp reads has been demonstrated in its paper : http://nar.oxfordjournals.org/conten...kt214.abstract

Subread is extremely fast. It only takes about 40 minutes to map 10 million pairs of 250bp reads using one thread. Subread can be downloaded from: http://subread.sourceforge.net

Cheers,
Wei

[opulcy97]

Hi Shi,

Thanks for such a good information. I will have a look at it.

Opulcy

[sparks]

Hi,

Novoalign Version 3 supports reads up to 1000bp

Colin

[AJenkins]

Hi,

Also, if you use Tophat2 which includes Bowtie you can use the parameters for whatever read length you may be interested in.

[FrankiB]

BWA mem

Hi everybody,
I got this error from my first try using bwa mem. Is it a problem of memory?

[M::main_mem] read 82646 sequences (10000166 bp)...
Segmentation fault (core dumped)


Thank you very much for your time

[mastal]

Segmentation faults are often a memory problem, although the number of sequences quoted in the error message doesn't seem like a lot.

How much memory does the genome index take up?

Do you have a small test data set that you can run, or try running the first 1000 reads from your sample to see if that works.

[AJenkins]

This normally means that the program is trying to access memory that isn't allocated/isn't allowed. This may be a deeper problem than just memory usage. Could have to do with accessing files/memory.

Are you using the latest version?

[FrankiB]

I'm using the 0.7.5a version and I want to map on the human genome g1k_V37. The index generated a couple of files and the biggest is around 3 gb.