program which can make a pair end to have equal number of sequence

[dejavu2010]

i have a PE100, the problem is some tiles of read 2 are corrupted and it caused un even number of reads in read 1 and read 2. Are there any programs which can match up sequences from both reads and just keep mathed ones for down stream analysis. Thanks

mike

[mgg]

I remember a nice contribution from kmcarr a while back which can probably help; search for thread 10392. (incidentally I can't recommend my own contribution to that thread - it is hideously slow)

best

m

[kmcarr]

Originally Posted by mgg

I remember a nice contribution from kmcarr a while back which can probably help; search for thread 10392. (incidentally I can't recommend my own contribution to that thread - it is hideously slow)

best

m

m,

Thanks for the acknowledgement. Here is a link to the thread. If you go there you'll see that I just posted an update. Due to a limitation in cdbfasta my method will not work for large input fastq files. The only work-around at the moment is to split the input up into smaller chunks.

[Hobbe]

The trimmer Trimmomatic outputs files with intact pairs as well as files with single reads. It should be able to split your files in the way you want, as well as do trimming at the same time if you wish.

[azneto]

I wrote a script exactly to tackle that issue. You'll find a copy attached.
The script will output either an interleaved mate pair fastq or two fastq files. The unpaired reads will also be saved in a separate file. The script uses a regular expression to identify the ID, so let me know if you need help with that. It requires at least as much RAM as the size of the first file. Feel free to use it and let please let me know how can we improve it.
Adhemar

[sklages]

Maybe of interest as well, PairedreadFinder:

Usage: PairedreadFinder, Version 1.01. This tool takes two fasta/q files and looks for matching readnames in both files. [OPTION]...

-h, --help displays this help message
-v, --version return program version
-s1, --source1 input file 1
-s2, --source2 input file 2
-f, --format input file format
-t1, --target1 target file 1
-t2, --target2 target file 2
-n, --nr-threads nr of threads to use (default 1)
-is, --suffix-ignore nr of characters to ignore from the END of the readname (in case paired reads are named like /1 /2 it should be set to 2) (default 0)
-ip, --prefix-ignore nr of characters to ignore from the BEGINNING of the readname (in case paired reads are named like s_1.. s_2.. it should be set to 3) (default 0)

from FAR, http://sourceforge.net/apps/mediawik...itle=Main_Page

Sven

[dejavu2010]

hi azneto, how to setup regular expression like the following
@HWI-ST829:138071VACXX:1:1101:1131:2048 1:N:0:ATCACG.

Thanks.

[azneto]

Hi dejavu2010.

@HWI-ST829:138071VACXX:1:1101:1131:2048 1:N:0:ATCACG.

you can use: '^@(\S+)\s[1|2]\S+$'

Assuming that 1 and 2 will appear right after the space char.
'@' 'ID' 'space' '1or1' '...'

I'll add this example to the script.

[dejavu2010]

Hi

my process got killed

perl mergeShuffledFastqSeqs.pl -f1 2044-BH-1_1_sequence.txt -f2 2044-BH-1_2_sequence.txt -r '^@(\S+)\s[1|2]\S+$' -o 2044-BH-1 -t

Loading the first file...Killed

2044-BH-1_1_sequence.txt 18gb, the other one is 17gb. we have a server with 32 duel core cpus and 192gb mem. I wonder what could be the reason it got killed.

thx

[epistatic]

Picard has a FixMateInformation to "Ensure that all mate-pair information is in sync between each read and it's mate pair."
http://picard.sourceforge.net/comman...ateInformation

If you are in Galaxy this is implemented under Picard as: Paired Read Mate Fixer

[azneto]

Hi,
It most probably is a memory issue.
The script loads only the first file into the memory and starts to match with the entries in the second file. You'll have to monitor the memory usage ('top' or 'free -m').
I just ran a test and perl uses 220Gb RAM for two 33Gb fastq file.
Soon I'll start to search for alternative ways to handle memory using perl in order to improve the script. I'll let you know.
-Adhemar

[dejavu2010]

thx. everybody, i got it resolved

[dejavu2010]

i feel that Trimmomatic index your reads based on input order, not lane_position_... combination, i tested one un matched dataset, they can not handle it.

[SES]

Originally Posted by kmcarr

m,

Thanks for the acknowledgement. Here is a link to the thread. If you go there you'll see that I just posted an update. Due to a limitation in cdbfasta my method will not work for large input fastq files. The only work-around at the moment is to split the input up into smaller chunks.

The other work-around is to stick with fasta files (smaller index files). If you are mapping to a reference and the quality scores are important then this probably won't help, but for assembly it doesn't matter.

Originally Posted by sklages

Maybe of interest as well, PairedreadFinder:
from FAR, http://sourceforge.net/apps/mediawik...itle=Main_Page

Sven

Has anyone actually used this program and found it to work correctly? I gave it a try but found several bugs. First, it inserted random blank lines in the individual "paired" output files. Second, the individual "paired" files actually differed by more than 800 records with my data, which means that trying to interleave the two files causes them to get all out of order. Of course, this could be due to the data or the user, but I've successfully used other methods with the same data and the usage of the program is quite simple, so I'm a bit skeptical on this one.

[kga1978]

Hi All,

I have exactly this problem as well, but with fasta files. Anybody know of a program that will work with Fasta or could modify 'mergeShuffledFastqSeqs.pl' so it would work on that format as well?

Much appreciated.

[carojasq]

I'm not sure that this works for you, but anyway if you have left and right fasta files and want to detect the paired-end and the single-end reads you will found usefull this script.https://github.com/lexnederbragt/den...leave_pairs.py
Needs biopython installed. http://biopython.org/wiki/Biopython

[kga1978]

That would probably have done the job, but unfortunately we don't have Biopython installed on our cluster so I can't run it (same with Bioperl).

What I am looking for is basically a standalone script that would do the trick - like the one for Fastq files, which works well.

[bmtb]

I can't get the mergeShuffledFastqSeqs.pl script to work with my data. I have two shuffled paired read files (one is 63GB in size and the other is 35GB). When I submit the job via batch script, it gets killed and I'm left with the following message: swap rate due to memory oversubscription is too high.
I've allocated 512GB of memory for this run so I don't think it has to do with that. Also, i've tried predefining the hash table size in the merge...pl script to be between 4-100 billion but this hasn't worked. Anyone have any ideas?

Thanks,
bmtb

Originally Posted by azneto

Hi,
It most probably is a memory issue.
The script loads only the first file into the memory and starts to match with the entries in the second file. You'll have to monitor the memory usage ('top' or 'free -m').
I just ran a test and perl uses 220Gb RAM for two 33Gb fastq file.
Soon I'll start to search for alternative ways to handle memory using perl in order to improve the script. I'll let you know.
-Adhemar

[azneto]

Hi bmtb,
Sorry it took me so long to reply.
The version of the script you have uses 40x the size of the f1 file.
I've just attached a version that uses about 6x.
So, if you use the 35GB file as f1 you should be able to run it this time.
Please let me know if it worked.
Perl hashes are really memory consuming structures and we're studing alternatives.
Best,
Adhemar

[bmtb]

Hi azneto,

I guess I should have posted this earlier, but I actually got your first script to work by increasing the memory allocation. Thanks for the updated script though.

Cheers,
bmtb