MaSuRCA failed to extract the best long reads

Question

[jon.brate]

I am trying to run a hybrid assembly of Illumina and ONT reads using MaSuRCA. I am running on a grid with a slurm system installed and I got a few errors during the run, but which did not stop the assembler. But finally the job stopped with the message failed to extract the best long reads. Anyone knows what's going on here?

Edit: The ONT reads are error corrected by LoRDEC and in fasta format. Perhaps they need to be in the ONT fast5 format?

Output:
Starting job 20198371 ("MaSuRCA") on c60-3 at Mon Feb 19 11:26:01 CET 2018
[Mon Feb 19 11:26:03 CET 2018] Processing pe library reads
[Mon Feb 19 11:26:06 CET 2018] Average PE read length 250
[Mon Feb 19 11:26:08 CET 2018] Using kmer size of 151 for the graph
[Mon Feb 19 11:26:10 CET 2018] MIN_Q_CHAR: 33
WARNING: JF_SIZE set too low, increasing JF_SIZE to at least 9282338590, this automatic increase may be not enough!
[Mon Feb 19 11:26:10 CET 2018] Creating mer database for Quorum
[Mon Feb 19 16:38:48 CET 2018] Error correct PE.
[Wed Feb 21 13:34:19 CET 2018] Estimated genome size: 375645933
[Wed Feb 21 13:34:19 CET 2018] Creating k-unitigs with k=151
ln: creating symbolic link `guillaumeKUnitigsAtLeast32bases_all.jump.fasta': File exists
[Fri Feb 23 06:24:03 CET 2018] Computing super reads from PE
slurmstepd: error: _get_pss: ferror() indicates error on file /proc/58696/smaps
slurmstepd: error: _get_pss: ferror() indicates error on file /proc/13419/smaps
[Mon Feb 26 00:25:57 CET 2018] Using linking mates
Using CABOG from is /cluster/home/jonbra/MaSuRCA-3.2.5/bin/../CA8/Linux-amd64/bin
/cluster/home/jonbra/MaSuRCA-3.2.5/bin/mega_reads_assemble_cluster.sh: line 127: [: -lt: unary operator expected
/cluster/home/jonbra/MaSuRCA-3.2.5/bin/mega_reads_assemble_cluster.sh: line 128: [: -gt: unary operator expected
Running mega-reads correction/assembly
Using mer size 15 for mapping, B=15, d=0.02
Estimated Genome Size 375645933
Estimated Ploidy
Using 16 threads
Output prefix mr.41.15.15.0.02
Using 30x of the longest ONT reads
failed to extract the best long read

Answer 1

[colindaven]

In my experience Lordec tends to mess up the long reads during correction by shortening them.

I would do it like this

1. canu
2. take corrected reads and assemble with masurca (if you want) OR
3. get canu assembly
4. polish assembly eg racon? nanopolish or medaka
5. further polish with pilon ?

[jon.brate]

Thanks for your suggestions. I'm also running Canu in parallel so I will see once it' finished.

[Kundan Kumar]

I am also getting the same error, If anyone know what's going on exactly??

Answer 2

[Kundan Kumar]

do fresh start from making the configuration file to running ./assemble.sh, it will work.. cheers

[sdoore]

I have been having this problem too, but a fresh start didn't seem to help. Any other suggestions for dealing with this error?