Mate pair de novo assembly

[AdaLovelace]

Hi everyone, i am having some problems with a de novo assembly. I have two MatePair files, one from Illumina MiSeQ and the other from IonTorrent. My problem is first how do i prepare the data before the assembly, and then, which assembler is better? (I have been working with Ray, MIRA, Newbler and SPAdes, these last two gave much better results in other data) In Illumina matepair, i tried to work with NxTrim but i need two files as an input and i do not know if a complementary file of the original will work, then i tried Trimmomatic too, and used these files as entrey (the original and the complementary i created) but i'm not sure if the output files will work for the assembler... Anyway, has anyone has any experience working with these kind of data? Which pipeline do you recommend? Thanks in advance.

[GenoMax]

For reference cross-posted: https://www.biostars.org/p/191400/

It is ok to cross-post on SeqAnswers.

[Markiyan]

Try sff2phd (from consed contrib section) on ion's sff / miseq_trimmer on fastq....

First you need to find out what are the naming rules for specifying (mate) pairs for your assembler of choice.
Usually most assemblers are happy with [template_name]/1 for forward read and [template_name]/2 for reverse read name in fastq format. Some may want interleaved files and some may want forward/reverse reads in the separate files.

Than we need to prepare the data.
In any case the matepairs:

>read(template)_name
[seq/2C][linker][seq/1]

have to be split:
>read/1
seq/1
>read/2
seq/2

Than for SFF files (including iontorrent):
get my sff2phd tool from the contrib section of the consed, update the linker sequence to the iontorrent's paired end one and run it (you can disable phd files generation, unless you would like to use phredPhrap assembly pipeline) and save fastq files.

sff2phd produces fastq files with offset Q+64, while miseq_trimmer uses a standard fastq offset Q+31.

For illumina files:
1. run the input through the FLASH
2. give an attached matepair splitter a try (at least on the overlapping ends file).

fastq_miseq_trimmer [ext_frags.fastq] -base=[(filename_base).mpe1] -out1=[mpe1_1.fastq] -outlq=[mpe1_se.fastq] -out2=[mpe1_2.fastq] -mpestd -minl=32 -noclip -ends=3

PS: If any improvements/fixes to the tools are required, feel free to contact me.
PPS: Please can you post the linker sequence for the iontorrent PE reads, so I can add it to my tool.