search

Nextomics / NextDenovo

Fast and accurate de novo assembler for long reads
GNU General Public License v3.0
350 stars 52 forks source link

NextDeNovo running for >4days with no expected outputs #210

Closed RominaSSBatista closed 1 week ago

RominaSSBatista commented 2 weeks ago

Question or Expected behavior I am trying to use NextDenovo to assembly a genome for a non-human primate (genome size expected = 2.7g). Data generated from frozen tissue (storage >10yrs) P2Solo --> 35x coverage. Trying to assembly after run guppy6.3.8-gpu for base calling.
Job running for ca. 4 days. It generated the following:

tree
.
├── 01.raw_align
│   ├── 01.db_split.sh
│   └── 01.db_split.sh.work
│       └── db_split1
│           └── nextDenovo.sh
├── 02.cns_align
├── 03.ctg_graph
├── input.fofn
├── run.cfg

It seems the job is not generating the expected output. I wonder If I need to change more parameters from my run.cfg (see below).

Operating system Which operating system and version are you using?

LSB Version:    :core-4.1-amd64:core-4.1-noarch:cxx-4.1-amd64:cxx-4.1-noarch:desktop-4.1-amd64:desktop-4.1-noarch:languages-4.1-amd64:languages-4.1-noarch:printing-4.1-amd64:printing-4.1-noarch
Distributor ID: CentOS
Description:    CentOS Linux release 7.9.2009 (Core)
Release:    7.9.2009
Codename:   Core

GCC What version of GCC are you using?

cc (GCC) 4.8.5 20150623 (Red Hat 4.8.5-44)

Python What version of Python are you using?

Python 3.8.5

NextDenovo What version of NextDenovo are you using?

nextDenovo 2.5.2

Additional context (Optional) After reading many issues from this repo I built the following run.cfg

[General]
job_type = slurm # here I use SLURM to manage jobs
job_prefix = nextDenovo
task = all # 'all', 'correct', 'assemble'
rewrite = yes # yes/no
deltmp = yes
rerun = 
parallel_jobs = 12 # number of tasks used to run in parallel
input_type = raw
read_type = ont # clr, ont, hifi
input_fofn = input.fofn
workdir = path-to-/NextDeNovo_Assembly
submit = sbatch -p long-serial --cpus-per-task=24 --mem-per-cpu=5g -o path-to-/nextdenovo.out -e path-to-nextdenovo.err path-to-my-script/assembly_NextDeNovo_test.sh

[correct_option]
read_cutoff = 10k
seed_cutoff = 20k
genome_size = 2.7g #estimated genome for Titi-Monkeys
blocksize = 3g   
seed_cutfiles = 10 
sort_options = -m 20g -t 30 -k 40
minimap2_options_raw = -x ava-pb -t 80
pa_correction = 12 # number of corrected tasks used to run in parallel, each corrected task requires ~TOTAL_INPUT_BASES/4 bytes of memory usage.
correction_options = -p 12

[assemble_option]
random_round = 50
minimap2_options_cns = -t 12 -k 31 -w 17
minimap2_options_map = -t 12
nextgraph_options = -a 1

I would appreciate any feedback, Best Regards Romina Batista

moold commented 2 weeks ago

Could you paste the log content to here?

RominaSSBatista commented 2 weeks ago

Hi @moold, here is the .log file from this run. Thank you in advance for your time.

pid45690.log.info.txt

Best, Romina

moold commented 2 weeks ago

Could you paste the content of file /gws/nopw/j04/rotcotm/rominab/projects/NextDeNovo_Assembly/01.raw_align/01.db_split.sh.work/db_split1/nextDenovo.sh.e to here?

RominaSSBatista commented 2 weeks ago

Could you paste the content of file /gws/nopw/j04/rotcotm/rominab/projects/NextDeNovo_Assembly/01.raw_align/01.db_split.sh.work/db_split1/nextDenovo.sh.e to here?

#!/bin/bash
set -xveo pipefail
hostname
cd /gws/nopw/j04/rotcotm/rominab/projects/NextDeNovo_Assembly/01.raw_align/01.db_split.sh.work/db_split1
( time  /home/users/rominab/softwares/NextDenovo/bin/seq_dump -f 10k -s 20000 -b 3g -n 12 -d /gws/nopw/j04/rotcotm/rominab/projects/NextDeNovo_Assembly/01.raw_align /gws/nopw/j04/rotcotm/rominab/projects/NextDeNovo_Assembly/input.fofn )
touch /gws/nopw/j04/rotcotm/rominab/projects/NextDeNovo_Assembly/01.raw_align/01.db_split.sh.work/db_split1/nextDenovo.sh.done
moold commented 2 weeks ago

nextDenovo.sh.e, not nextDenovo.sh

RominaSSBatista commented 2 weeks ago

I am afraid that nextDenovo.sh.e file was not generated. Below I pasted the output if I run the tree command from the output directory:

$ tree
.
├── 01.raw_align
│   ├── 01.db_split.sh
│   └── 01.db_split.sh.work
│       └── db_split1
│           └── nextDenovo.sh
├── 02.cns_align
├── 03.ctg_graph
├── input.fofn
├── run2.cfg
└── run.cfg

5 directories, 5 files

Does that mean NexDenovo did not even run anything?

I appreciate your thought about this issue.

moold commented 2 weeks ago

Yes, have you tried clearing the directory and running again? Also, you can set job_type = local and try again.

RominaSSBatista commented 2 weeks ago

I just made it:

tree
.
├── 01.raw_align
│   ├── 01.db_split.sh
│   ├── 01.db_split.sh.work
│   │   └── db_split1
│   │       ├── nextDenovo.sh
│   │       ├── nextDenovo.sh.e
│   │       └── nextDenovo.sh.o
│   ├── input.part.001.2bit
│   ├── input.seed.001.2bit
│   ├── input.seed.002.2bit
│   ├── input.seed.003.2bit
│   ├── input.seed.004.2bit
│   ├── input.seed.005.2bit
│   ├── input.seed.006.2bit
│   ├── input.seed.007.2bit
│   ├── input.seed.008.2bit
│   ├── input.seed.009.2bit
│   ├── input.seed.010.2bit
│   ├── input.seed.011.2bit
│   └── input.seed.012.2bit
├── 02.cns_align
├── 03.ctg_graph
├── input.fofn
├── run2.cfg
└── run.cfg

I am also sending attached the nextDenovo.sh.e file

RominaSSBatista commented 2 weeks ago

Dear @moold, my job seems to be working better now, getting expected outputs. I just wonder how long it would take for this assembly to finish (as mentioned I have 35x coverage (after Guppy base call), expected genome size=2.7Gb). I have up to 7 days on the queue I am currently running, and had allocated -n24 --mem-per-cpu=5G. I am wondering if these are reasonable infrastructure. Best, Romina

RominaSSBatista commented 2 weeks ago

In approximately 4h and so, my my job got killed, and I am sending the log file produced here pid85429.log.info.txt.

I am also sending here the .sh.e files that are mentioned at the end of the log sort_align02_nextDenovo.sh.e.txt

sort_align08_nextDenovo.sh.e.txt

sort_align12_nextDenovo.sh.e.txt

from my 

$ tree

217 directories, 1171 files

I would appreciate if you could share your thoughts again.

Best, Romina Batista

moold commented 2 weeks ago

The computer RAM is not enough, so it killed all subtaskes. You can run these failed tasks manually using command: sh sort_align02_nextDenovo.sh

After you manually run all failed tasks, you can continue to run the main task.

RominaSSBatista commented 2 weeks ago

The computer RAM is not enough, so it killed all subtaskes. You can run these failed tasks manually using command: sh sort_align02_nextDenovo.sh

After you manually run all failed tasks, you can continue to run the main task.

Hi, @moold, thank you for your reply. I did run all failed tasks, interactively on the HPC I am working, it was very straight forward and fast, from my terminal. I am wondering now how I should continue the main task. I have used the following sbatch script to submit my job: 

#!/bin/bash

###########################################
############## Romina Batista #################
############  r.d.s.d.s.batista@salford.ac.uk ########
################### 2024 ###################

###########################################
### 3. Run NextDenovo for a de novo assembly
### https://nextdenovo.readthedocs.io/en/latest/QSTART.html
### De Novo Assembly Project from Nanopore P2Solo 35x cov
#############################################
# I have used this issue on github to be able to 
# tunning some parameters
# https://github.com/Nextomics/NextDenovo/issues/170
# and I have been discussing the use with the developer here:
# https://github.com/Nextomics/NextDenovo/issues/210
##############################################

#SBATCH --partition=long-serial
#SBATCH --time=168:00:00
#SBATCH -n 24
#SBATCH --mem-per-cpu=5G
#SBATCH --job-name=NextDeNovo
#SBATCH --error=/gws/nopw/j04/rotcotm/rominab/projects/error/assemb_nextdenov_9jun24.err
#SBATCH --output=/gws/nopw/j04/rotcotm/rominab/projects/error/assemb_nextdenov_9jun24.out

source /home/users/rominab/VirEnvPy3/bin/activate

NextDeNovo_dir='/home/users/rominab/software/NextDenovo'
config_dir='/gws/nopw/j04/rotcotm/rominab/projects/NextDeNovo_Assembly'

$NextDeNovo_dir/nextDenovo  $config_dir/run2.cfg

where run2.cfg file looks like below: 

[General]
job_type = local # I set up as local, as a recommendation from the developer
job_prefix = nextDenovo
task = all # 'all', 'correct', 'assemble'
rewrite = yes # yes/no
deltmp = yes
rerun = 
parallel_jobs = 12 # number of tasks used to run in parallel
input_type = raw
read_type = ont # clr, ont, hifi
input_fofn = /gws/nopw/j04/rotcotm/rominab/projects/NextDeNovo_Assembly/input.fofn
workdir = /gws/nopw/j04/rotcotm/rominab/projects/NextDeNovo_Assembly
submit = sbatch -p long-serial --cpus-per-task=24 --mem-per-cpu=5g -o /gws/nopw/j04/rotcotm/rominab/projects/error/nextdenovo.out -e /gws/nopw/j04/rotcotm/rominab/projects/error/nextdenovo.out /gws/nopw/j04/rotcotm/rominab/projects/scripts/assembly_NextDeNovo_test.sh

[correct_option]
read_cutoff = 10k
seed_cutoff = 20k
genome_size = 2.7g #estimated genome for Titi-Monkeys
blocksize = 3g   
seed_cutfiles = 10 
sort_options = -m 20g -t 30 -k 40
minimap2_options_raw = -x ava-pb -t 80
pa_correction = 12 # number of corrected tasks used to run in parallel, each corrected task requires ~TOTAL_INPUT_BASES/4 bytes of memory usage.
correction_options = -p 12

[assemble_option]
random_round = 50
minimap2_options_cns = -t 12 -k 31 -w 17
minimap2_options_map = -t 12
nextgraph_options = -a 1

My main question here is if I use the same sbatch file to submit, does NextDenovo be able to recognise where it stopped? Should I adjust any line on my run2.cfg file? Maybe task = ??? # 'all', 'correct', 'assemble'

rewrite = ??? # yes/no

Many thanks for all your attention and time with this issue.

Best, Romina

moold commented 2 weeks ago

Just run the same command/sbatch file again, see here.

RominaSSBatista commented 2 weeks ago

Thank you! It looks like NextDeNovo crashed again in this next step: pid193698.log.info

and here is one one the sh.e file mentioned at the end of the log file: nextDenovo.sh.e

I wonder if I can fix this as I did before, by running manually those reported at the end of the log file?

Thank you, Romina

moold commented 2 weeks ago

Yes

RominaSSBatista commented 2 weeks ago

Dear @moold, as mentioned previously I run the 1st one manually and I got this at the end of stout msg on my terminal: 

[909 INFO] 2024-06-19 05:09:41 Start a cns worker in 909 from parent 31397
[945 INFO] 2024-06-19 05:09:45 Start a cns worker in 945 from parent 31397
[955 INFO] 2024-06-19 05:09:50 Start a cns worker in 955 from parent 31397
[962 INFO] 2024-06-19 05:09:53 Start a cns worker in 962 from parent 31397
[978 INFO] 2024-06-19 05:09:57 Start a cns worker in 978 from parent 31397
[985 INFO] 2024-06-19 05:09:58 Start a cns worker in 985 from parent 31397
[993 INFO] 2024-06-19 05:09:59 Start a cns worker in 993 from parent 31397
[1009 INFO] 2024-06-19 05:10:03 Start a cns worker in 1009 from parent 31397
[1025 INFO] 2024-06-19 05:10:07 Start a cns worker in 1025 from parent 31397
[1029 INFO] 2024-06-19 05:10:08 Start a cns worker in 1029 from parent 31397
[1050 INFO] 2024-06-19 05:10:11 Start a cns worker in 1050 from parent 31397
[31612 INFO] 2024-06-19 05:10:15 Start a cns worker in 31612 from parent 31397
[1060 INFO] 2024-06-19 05:10:15 Start a cns worker in 1060 from parent 31397
[1063 INFO] 2024-06-19 05:10:16 Start a cns worker in 1063 from parent 31397
[1070 INFO] 2024-06-19 05:10:16 Start a cns worker in 1070 from parent 31397
[1074 INFO] 2024-06-19 05:10:17 Start a cns worker in 1074 from parent 31397
[1083 INFO] 2024-06-19 05:10:20 Start a cns worker in 1083 from parent 31397
[1084 INFO] 2024-06-19 05:10:20 Start a cns worker in 1084 from parent 31397
[1092 INFO] 2024-06-19 05:10:24 Start a cns worker in 1092 from parent 31397
[1093 INFO] 2024-06-19 05:10:24 Start a cns worker in 1093 from parent 31397
[2109 INFO] 2024-06-19 05:12:57 Start a cns worker in 2109 from parent 31397
[2111 INFO] 2024-06-19 05:12:57 Start a cns worker in 2111 from parent 31397
nextDenovo.sh: line 5: 31397 Killed                  /home/users/rominab/VirEnvPy3/bin/python /home/users/rominab/softwares/NextDenovo/lib/nextcorrect.py -f /gws/nopw/j04/rotcotm/rominab/projects/NextDeNovo_Assembly/02.cns_align//01.seed_cns.input.idxs -i /gws/nopw/j04/rotcotm/rominab/projects/NextDeNovo_Assembly/01.raw_align/03.sort_align.sh.work/sort_align01/input.seed.001.sorted.ovl -p 12 -min_len_seed 10000 -max_lq_length 1000 -r ont -o cns.fasta

indeed the cns.fasta is not empty, so I am assuming it finished well. Should I worry about this "killed" message? Below the files generated at this step, por this specific task and the size of the files: 

[rominab@sci2 seed_cns01]$ tree
.
├── cns.fasta
├── cns.fasta.idx
├── nextDenovo.sh
├── nextDenovo.sh.e
└── nextDenovo.sh.o

0 directories, 5 files

and:

[rominab@sci2 seed_cns01]$ du -sh *
775M    cns.fasta
795K    cns.fasta.idx
4.0K    nextDenovo.sh
70K nextDenovo.sh.e
4.0K    nextDenovo.sh.o

I would appreciate, again, if you share your thoughts. Thank you, Romina

moold commented 2 weeks ago

Yes, if there is a new file "nextDenovo.sh.done", it means this task is completed. The computer RAM is not enough, so you can try to reduce correction_options = -p 12 to correction_options = -p 5, if this doesn't work, you need change a computer with more RAM.

RominaSSBatista commented 1 week ago

Dear @moold, I manage to run those task one by one using --mem=500G on my slurm scripts (I used high mem here with the intention of not getting my job killed by the HPC). I did not change correction_options = -p 12 to correction_options = -p 5. Adjust mem was enough to solve.

The next step also got stuck, and again, I manually finished all. It is very useful and easy, from the log file that your software generates, to find the next step to be fixed. I would strongly recommend adding this to your documentation. Although, somehow if I submit to the SLURM system it would not behave as good as running those small tasks interactively. Regardless, I finally managed to finish my assembly. Thank you a lot for your support during this week.

Just before closing this issue, I have a final question: Is there any way to tweak the parameters to improve N50? I am adding bellow the results from my stats and I got very worried by the poor draft genome I was able to generate using my data from Nanopore using NextDeNovo:

N50 of 0.1 Mb from ca. 1K contigs is way far from 2Mb I managed to recover running wtdbg2, although wtdbg2 used ca.20K contigs. The latter software is known to generate "short" N50. I was amazed by the fact NextDeNovo generated even worse N50. 

[34217 INFO] 2024-06-19 22:41:08 asm stat:
[34217 INFO] 2024-06-19 22:41:08 
Type           Length (bp)            Count (#)
N10               244613                 139
N20               207530                 328
N30               182044                 545
N40               165150                 789
N50               149992                1058
N60               134362                1355
N70               118331                1689
N80               101887                2073
N90                84225                2528

Min.               24628                   -
Max.              844975                   -
Ave.              135499                   -
Total          422080978                3115

I would appreciate any thoughts you could share about how to improve N50 prior to run NextPolish to polish, since it is the next step I will take.

Best, Romina

RominaSSBatista commented 1 week ago

A follow up from my previous comment is that I will try: 

nextgraph_options = -a 1 -q 10

and compare my results from my previous run that was set as:

nextgraph_options = -a 1

Following FAQ

Best, Romina

moold commented 1 week ago

Don't set seed_cutoff, let Nextdenovo calculate it automatically. I'm not sure if it will improve the assemble N50, but you can try it.

RominaSSBatista commented 1 week ago

Currently running both:

parameter1

[General]
job_type = local # I set up as local, as a recommendation from the developer
job_prefix = nextDenovo
task = all # 'all', 'correct', 'assemble'
rewrite = yes # yes/no
deltmp = yes
rerun = 
parallel_jobs = 12 # number of tasks used to run in parallel
input_type = raw
read_type = ont # clr, ont, hifi
input_fofn = /gws/nopw/j04/rotcotm/rominab/projects/NextDeNovo_Assembly_2/input.fofn
workdir = /gws/nopw/j04/rotcotm/rominab/projects/NextDeNovo_Assembly_2
submit = sbatch -p high-mem --cpus-per-task=24 --mem=256g -o /gws/nopw/j04/rotcotm/rominab/projects/6_Pgrovesi_Nanopore/error/Assemb_nextgraph.out -e /gws/nopw/j04/rotcotm/rominab/projects/6_Pgrovesi_Nanopore/error/Assemb_nextgraph.err /gws/nopw/j04/rotcotm/rominab/projects/6_Pgrovesi_Nanopore/scripts/assembly_NextDeNovo_nextgraph.sh

[correct_option]
read_cutoff = 10k
seed_cutoff = 20k
genome_size = 2.7g #estimated genome for Titi-Monkeys
blocksize = 3g   
seed_cutfiles = 10 
sort_options = -m 20g -t 30 -k 40
minimap2_options_raw = -x ava-pb -t 80
pa_correction = 12 # number of corrected tasks used to run in parallel, each corrected task requires ~TOTAL_INPUT_BASES/4 bytes of memory usage.
correction_options = -p 12

[assemble_option]
random_round = 50
minimap2_options_cns = -t 12 -k 31 -w 17
minimap2_options_map = -t 12
nextgraph_options = -a 1 -q 10

&

parameter2

[General]
job_type = local # I set up as local, as a recommendation from the developer
job_prefix = nextDenovo
task = all # 'all', 'correct', 'assemble'
rewrite = yes # yes/no
deltmp = yes
rerun = 
parallel_jobs = 12 # number of tasks used to run in parallel
input_type = raw
read_type = ont # clr, ont, hifi
input_fofn = /gws/nopw/j04/rotcotm/rominab/projects/NextDeNovo_Assembly_2/input.fofn
workdir = /gws/nopw/j04/rotcotm/rominab/projects/NextDeNovo_Assembly_2
submit = sbatch -p high-mem --cpus-per-task=24 --mem=256g -o /gws/nopw/j04/rotcotm/rominab/projects/6_Pgrovesi_Nanopore/error/Assemb_nextgraph.out -e /gws/nopw/j04/rotcotm/rominab/projects/6_Pgrovesi_Nanopore/error/Assemb_nextgraph.err /gws/nopw/j04/rotcotm/rominab/projects/6_Pgrovesi_Nanopore/scripts/assembly_NextDeNovo_nextgraph.sh

[correct_option]
read_cutoff = 10k
genome_size = 2.7g #estimated genome for Titi-Monkeys
blocksize = 3g   
seed_cutfiles = 10 
sort_options = -m 20g -t 30 -k 40
minimap2_options_raw = -x ava-pb -t 80
pa_correction = 12 # number of corrected tasks used to run in parallel, each corrected task requires ~TOTAL_INPUT_BASES/4 bytes of memory usage.
correction_options = -p 12

[assemble_option]
random_round = 50
minimap2_options_cns = -t 12 -k 31 -w 17
minimap2_options_map = -t 12
nextgraph_options = -a 1 -q 10

I will let you know what perfomed better as soon as it finishes.

All your comments are much appreciated @moold!

Romina

RominaSSBatista commented 1 week ago

Currently running both:

parameter1

[General]
job_type = local # I set up as local, as a recommendation from the developer
job_prefix = nextDenovo
task = all # 'all', 'correct', 'assemble'
rewrite = yes # yes/no
deltmp = yes
rerun = 
parallel_jobs = 12 # number of tasks used to run in parallel
input_type = raw
read_type = ont # clr, ont, hifi
input_fofn = /gws/nopw/j04/rotcotm/rominab/projects/NextDeNovo_Assembly_2/input.fofn
workdir = /gws/nopw/j04/rotcotm/rominab/projects/NextDeNovo_Assembly_2
submit = sbatch -p high-mem --cpus-per-task=24 --mem=256g -o /gws/nopw/j04/rotcotm/rominab/projects/6_Pgrovesi_Nanopore/error/Assemb_nextgraph.out -e /gws/nopw/j04/rotcotm/rominab/projects/6_Pgrovesi_Nanopore/error/Assemb_nextgraph.err /gws/nopw/j04/rotcotm/rominab/projects/6_Pgrovesi_Nanopore/scripts/assembly_NextDeNovo_nextgraph.sh

[correct_option]
read_cutoff = 10k
seed_cutoff = 20k
genome_size = 2.7g #estimated genome for Titi-Monkeys
blocksize = 3g   
seed_cutfiles = 10 
sort_options = -m 20g -t 30 -k 40
minimap2_options_raw = -x ava-pb -t 80
pa_correction = 12 # number of corrected tasks used to run in parallel, each corrected task requires ~TOTAL_INPUT_BASES/4 bytes of memory usage.
correction_options = -p 12

[assemble_option]
random_round = 50
minimap2_options_cns = -t 12 -k 31 -w 17
minimap2_options_map = -t 12
nextgraph_options = -a 1 -q 10

Results - Stats for parameter1

Type           Length (bp)            Count (#)
N10               255304                 320
N20               201338                 796
N30               171770                1370
N40               151082                2032
N50               135789                2776
N60               122250                3602
N70               110702                4518
N80                98515                5533
N90                84163                6698

Min.               12547                   -
Max.             1393551                   -
Ave.              129917                   -
Total         1063761647                8188

&

parameter2

[General]
job_type = local # I set up as local, as a recommendation from the developer
job_prefix = nextDenovo
task = all # 'all', 'correct', 'assemble'
rewrite = yes # yes/no
deltmp = yes
rerun = 
parallel_jobs = 12 # number of tasks used to run in parallel
input_type = raw
read_type = ont # clr, ont, hifi
input_fofn = /gws/nopw/j04/rotcotm/rominab/projects/NextDeNovo_Assembly_2/input.fofn
workdir = /gws/nopw/j04/rotcotm/rominab/projects/NextDeNovo_Assembly_2
submit = sbatch -p high-mem --cpus-per-task=24 --mem=256g -o /gws/nopw/j04/rotcotm/rominab/projects/6_Pgrovesi_Nanopore/error/Assemb_nextgraph.out -e /gws/nopw/j04/rotcotm/rominab/projects/6_Pgrovesi_Nanopore/error/Assemb_nextgraph.err /gws/nopw/j04/rotcotm/rominab/projects/6_Pgrovesi_Nanopore/scripts/assembly_NextDeNovo_nextgraph.sh

[correct_option]
read_cutoff = 10k
genome_size = 2.7g #estimated genome for Titi-Monkeys
blocksize = 3g   
seed_cutfiles = 10 
sort_options = -m 20g -t 30 -k 40
minimap2_options_raw = -x ava-pb -t 80
pa_correction = 12 # number of corrected tasks used to run in parallel, each corrected task requires ~TOTAL_INPUT_BASES/4 bytes of memory usage.
correction_options = -p 12

[assemble_option]
random_round = 50
minimap2_options_cns = -t 12 -k 31 -w 17
minimap2_options_map = -t 12
nextgraph_options = -a 1 -q 10

Results - Stats for parameter2

Type           Length (bp)            Count (#)
N10              1899231                  96
N20              1302148                 252
N30               973543                 469
N40               732325                 755
N50               546282                1136
N60               404893                1653
N70               285287                2365
N80               195137                3383
N90               125235                4923

Min.               16174                   -
Max.             5437132                   -
Ave.              311254                   -
Total         2410043747                7743

:heavy_exclamation_mark: In summary parameter2 performed better among all trials I did, but still not as I was expecting, hopefully I can improving that by polishing it.

:pushpin: Both assemblies finished in about 7h (very fast!) after setting --mem=256G.

Thank you very much @moold for your time and support, I will now close this issue.

Romina