floflooo

Hi Muhammet, First, thing, the default -abundance_model is uniform, so if your goal is to produce an uniform distribution of all reference sequences, you can omit the -abundance_file parameter. Second, the default -total_reads is 100, so with such as small number, you may not have exactly 10 reads from each of your 10 references, but close to that (since this process is based on probabilities). Hope this helps! Best, Florent

Hi Laetitia, I finally had the time to have a look at your example. The problem is that you added several sources of errors in your command. Specifically, the Balzer homopolymer model is the one that causes the extra indels you observe. Since you want to model Illumina reads anyway, omit any homopolymer model (these models are more suited for 454 reads). And let me re-iterate what I said in a previous email. If you run into issues, you need to find the smallest reproducible example (few reference...

Hi Guilherme, Grinder is not multi-threaded. However, you can run several Grinder processes simultaneously to take advantage of multiple processing cores. For example, you create a small Bash script that runs 1 Grinder process per chromosome. A note about Grinder speed. There can be a specific performance decrease when you use very large genomes (e.g. human chromosomes) or a very large number of smaller genomes, since all reference sequences are read and kept in memory simultaneously by Grinder....

Hi Aditi, Grinder was written to simulate sequencing errors, but you could likely indirectly use it to include genomic variant "errors". This is an approach I have not tested: 1/ Input your genomes one by one in Grinder and introduce some errors using the error model desired. For the read length parameter, you will need to provide the exact length of your genome to avoid it being truncated or excluded. 2/ After your variants are ready, concatenate them into a single FASTA/FASTQ file. You may need...

Mutations when no error model specified

Hi Chad, Have a look at the description of the reads: 1 reference=test1 position=1..99 2 reference=test1 position=complement(1..99) No errors were introduced. However, as you see, some of the reads were generated from the reference sequence itself and some others from its reverse-complement, as you would expect from random shotgun sequencing. If your use-case calls for unidirectional reads, have a look at the -unidirectional option. Cheers, Florent

Hi Mao, The Illumina model from Korbel assumes that the reads are 100bp. You will need to adjust the parameters if your reads are longer. I suggest you experiment in Excel with this equation (Korbel et al 2009) to keep the same error rate at the last position as you find for reads that are 100 bp: 3e-3 + 3.3e-8 * i^4 Best, Florent

Hi Mao, That's an interesting idea, but it is not supported by Grinder. Best, Florent On Wed, Jun 20, 2018, 20:29 Maoxuan Lin mlin18@users.sourceforge.net wrote: Hi Ginder, I am using Grinder to simulate paired-end reads using a set of reference sequences. My goal is to guarantee every reference sequence has been selected to simulate reads. When I increased the total number of reads to large enough, I achieved this goal. However, I could not find a general rule or specific parameters to achieve this;...