wtdbg get smaller genome assembly

[ruanjue]

wtdbg get smaller genome assembly

[ruanjue]

Actually, I have no experience on CANU->wtdbg. If you have no other better way, try use --aln-noskip . Also please have a look at wtdbg-1.2.8 --help.

[zkstewart]

From my experience using a few assemblers (Canu, HGAP4, MECAT, Miniasm, SMARTdenovo, WTDBG, and Flye) what you're seeing is normal. Specifically, for my species, both Canu and MECAT give greatly increased genome sizes (approx. 430-450Mb) relative to the other assemblers (approx. 270-300Mb).

From doing things like running Redundans on the genomes and BUSCO on my predicted gene models, both Canu and MECAT have a lot of redundancy, i.e., they're assembling the diploid strands individually, whereas most other assemblers are collapsing these regions (which is the behaviour that I actually want).

As an example, my Canu BUSCO score has 73.3% duplicated models, whereas WTDBG has only 0.9% (!). Things like this lead me to believe that WTDBG provides the most realistic assembly size out of the programs I've used (even though SMARTdenovo seems to give me the most balanced result for my species from what I've seen thus far).

I think if you do a similar analysis including Redundans and gene model prediction you'll be able to see if Canu's large genome size is real, or if WTDBG's is more accurate similar to what I've seen with my data.

Zac.

[ruanjue]

Thanks ZAC. 'Redundans and BUSCO' is really a good way to select one from various assemblies.

You also bring me a useful news about MECAT. I had tried it on c.elegans PacBio assembly, but got ./mecat.sh: line 7: 56374 Segmentation fault (core dumped) mecat2cns -i 0 -t 96 pairwise.pm.m4 wt.fa corrected.fasta, here, wt.fa was the input seuqences file. Do you by chance know about it?

Regards,
Jue

[tangerzhang]

Thanks, Zac and Jue. I will try what you did!

[zkstewart]

Hi Jue,

I'm not sure if I encountered that error with mecat2cns, it was a while ago that I performed the assembly. If it helps, this is the script I used for running the program which finished successfully. It might help to see how I formatted all of my program calls?

export HDF5_INCLUDE=/home/stewarz3/various_programs/hdf5/include
export HDF5_LIB=/home/stewarz3/various_programs/hdf5/lib

export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/stewarz3/various_programs/hdf5/lib
export PATH=/home/stewarz3/various_programs/MECAT/Linux-amd64/bin:$PATH
export PATH=/home/stewarz3/various_programs/DEXTRACTOR:$PATH

FILEDIR=/home/stewarz3/species_assembly/assembly_ready
FILENAME=species_subreads
PREFIX=species_mecat

CPUS=12
MEM=80
GENSIZE=300000000
COV=75

mecat2pw -j 0 -d $FILEDIR/${FILENAME}.fasta -o ${PREFIX}.fasta.pm.can -w wrk_dir -t $CPUS
mecat2cns -i 0 -t 16 ${PREFIX}.fasta.pm.can $FILEDIR/${FILENAME}.fasta corrected_species_filtered.fasta
extract_sequences corrected_species_filtered.fasta corrected_species_75x.fasta $GENSIZE $COV
mecat2canu -trim-assemble -p $PREFIX -d $PREFIX genomeSize=$GENSIZE ErrorRate=0.02 maxMemory=$MEM maxThreads=$CPUS useGrid=0 Overlapper=mecat2asmpw -pacbio-corrected corrected_species_75x.fasta.fasta

Otherwise, I used MECAT commit de87b0b4794a3270a1f5c2bc92c7ce15653574ab. Sorry, I'm not sure if I can help much beyond this.

[ruanjue]

ZAC, Thanks! I will try it again.