Developing microsatellites from pacbio reads

[anjama]

Disclaimer: everything I know about Pacbio has been learned over the process of the last few days. I'm basically the closest thing to a bioinformatician in our lab, so I'm helping some people figure out the software side of this.

Basically, their goal is to use pacbio reads to develop microsatellite markers for a bat species (probably has a genome size of 2-3 Gbp). I think they only ran a single cell, and received the raw hdx.h5 data files, as well as some filtered subread files in fastq and fasta formats.

I've been trying to understand the massive number of tools available and how they interact (SMRT analysis, CANU, MIRA, the Celera Assembler, etc). We don't really have anywhere near the data required to try a full de novo assembly, and for our purposes, it wouldn't be necessary anyway. I have the SMRT Analysis tools up and running on my computer, and so far the PreAssembler tool seems like the best bet. I've successfully run it using mostly default settings, but don't know enough about the tool to know if the results are reasonable. Part of the reason I'm doubting it is because I believe their target sequence length was 12000 bp, and it doesn't seem like the assembled reads are reflecting that. So if there is an issue, I'm not sure if it's incorrect understanding of how the software works on my part, improper parameter values, or the lab we sent our samples to be sequenced, which apparently other people have had issues with. I've attached the report files if anyone would be willing to give it a look over.

Alternatively, it seems like using the PreAssembler might be overcomplicating things. I've seen mention of CCS fasta files, and it seems like these might be adequate. However, it seems like most people receive them, yet we didn't receive any, and I'm having trouble figuring out which tool I can use to create them from the the raw data. If anyway could point me in the right direction for that, it would be appreciated. Thanks

[gconcepcion]

Disclaimer: everything I know about Pacbio has been learned over the process of the last few days. I'm basically the closest thing to a bioinformatician in our lab, so I'm helping some people figure out the software side of this.

Basically, their goal is to use pacbio reads to develop microsatellite markers for a bat species (probably has a genome size of 2-3 Gbp). I think they only ran a single cell, and received the raw hdx.h5 data files, as well as some filtered subread files in fastq and fasta formats.

I've been trying to understand the massive number of tools available and how they interact (SMRT analysis, CANU, MIRA, the Celera Assembler, etc). We don't really have anywhere near the data required to try a full de novo assembly, and for our purposes, it wouldn't be necessary anyway. I have the SMRT Analysis tools up and running on my computer, and so far the PreAssembler tool seems like the best bet. I've successfully run it using mostly default settings, but don't know enough about the tool to know if the results are reasonable. Part of the reason I'm doubting it is because I believe their target sequence length was 12000 bp, and it doesn't seem like the assembled reads are reflecting that. So if there is an issue, I'm not sure if it's incorrect understanding of how the software works on my part, improper parameter values, or the lab we sent our samples to be sequenced, which apparently other people have had issues with. I've attached the report files if anyone would be willing to give it a look over.

Alternatively, it seems like using the PreAssembler might be overcomplicating things. I've seen mention of CCS fasta files, and it seems like these might be adequate. However, it seems like most people receive them, yet we didn't receive any, and I'm having trouble figuring out which tool I can use to create them from the the raw data. If anyway could point me in the right direction for that, it would be appreciated. Thanks

Hi Anjama -

To screen for microsatellites I would suggest a starting point of either Pre-assembled reads or CCS reads as you mentioned above. Pre-assembled reads are probably best, however, if you're attempting pre-assembly with less than 1X coverage of the genome, you are going to get very poor pre-assembly results. There will simply not be enough overlaps to generate a high quality pre-assembly.

In your situation, generating the single molecule consensus sequences (CCS reads) and subsequently screening those reads for STRs is more than likely going to be your best option in this circumstance.

In the past, the CCS reads were generated during the sequencing process. That workflow was moved downstream to secondary analysis, and in SMRTAnalysis V2.3.0 you should be able to use the "RS_ReadsOfInsert.1" protocol to generate your CCS fasta file.

Are you running smrtanalysis from the command line, or from the GUI?