SOAPdenovo and FASTA files - Newbie question

[Gregory Peterson]

I'm a newbie when it comes to NGS bioinformatics. I'm testing SOAPdenovo with a small FASTA file (3300 kb). The FASTA file is an unpaired set of exported reads viewed in Tablet but originally generated in a BWA alignment, i.e. I extracted a subset of reads from my MiSeq run based on a reference sequence for denovo assembly in SOAPdenovo. For an unknown reason SOAPdenovo fails to read the FASTA file. (see below)

Has anyone else seen this and is there a fix out there?

Version 2.04: released on July 13th, 2012
Compile Apr 25 2013 16:59:53

********************
Pregraph
********************

Parameters: pregraph -s soap-fasta.config -K 63 -R -o fastatest

In soap-fasta.config, 1 lib(s), maximum read length 260, maximum name length 256.

8 thread(s) initialized.
Import reads from file:
/home/fasta/1700-sorted-bam_140k.txt
--- 100000000th reads.
--- 200000000th reads.

...and on and on to "kill -9"

[Kennels]

Hi

What were your:

1. command line invocations
2. config file contents
3. fasta header description

[Gregory Peterson]

fasta, config, and command line

FASTA headers (original, modified 1, and modified 2):

>M00542:7:000000000:A3C86:1:2103:28825:15020 pos=33 len=89

or
>M00542_7_000000000-A3C86_1_1102_26169_15631_pos-125_len-90

or
>1101_11975_19083_pos-230_len-42




.config file (I've also hashed-out avg_ins and had asm_flags set to 1 or 3)


#maximal read length
max_rd_len=260
[LIB]
#average insert size
avg_ins=420
#if sequence needs to be reversed
reverse_seq=0
#in which part(s) the reads are used
asm_flags=1
#in which order the reads are used while scaffolding
rank=1
#fasta file (unpaired ends)
f=/home/fasta/1700-sorted_bam.fasta


Command line: (note that the executable was renamed to 'soapdenovo')

>soapdenovo all -s soap-fasta.config -K 63 -R -o testfasta 1>fastaass.log 2>fastaass.err

[Gregory Peterson]

There appears to be occasional astericks (*) in the Tablet data that caused the fault. Removing these seemed to do the trick.

Thanks for your time.

[Topulaneus-Hattum]

There appears to be occasional astericks (*) in the Tablet data that caused the fault. Removing these seemed to do the trick.

Gregory, where in your data did the asterisks appear? Among the nucleotides? Sequence names?

I am experiencing a similar problem with soap denovo. My input is a pair of gzipped fastq files, 100 million 150 bp reads in each. Soap denovo ran for 60 hours and was reporting "--- 18600000000th reads" before I killed it.

I extracted a small subset of the pairs (a couple thousand), and soap denovo ran fine for those.

An online search revealed your post of a similar problem. It sounds like I need to "fix" my reads to remove asterisks but I want to understand better what you actually changed.

Thanks.

[Gregory Peterson]

Tablet asterisks in nucleotide sequence

The asterisk symbols were located in the actual DNA nucleotide sequence and from what I was able to determine were being used as 'gap' fillers by Tablet (as opposed to dashes or periods). Removing them using grep or other find-and-replace software did the trick for me.

[Topulaneus-Hattum]

Thanks, Gary,

I think my case triggered the same problem, but for different reasons.

I think I must have experienced a transient corruption of my input files. I wasn't able to find any non ACGTN characters among my nucleotide sequences. Eventually I decided to just running the same thing a second time, and it ran without any problems.

We do occasionally have I/O issues on the machine I was running on, usually these manifest as something like a "network time out" being reported to standard error. If this is what caused the problem with SOAPdenovo, the manifestation is quite unusual, there was no such report to stderr.

Thanks for your response,
T. Hattum