I got some colonies that showed phenotype difference with my wild type strain. I have picked 9
randomly and did whole genome sequencing. Now I have assembled these 9 strains with contigs (not closed genome) and I would like to compare with wild type (closed genome). Does anybody have any experience/ideas on how could I find the mutation sites?
hi Xiyan Wang
i think you need to know ancestor first then you can show the different between your sequence (Query) and ancestor , so you can find the mutation sites .
For your task it would be more convenient to map sequenced reads on a reference genome instead of de novo assembly. Mapped reads are then being used for variant calling, i.e. for scoring all differences from the reference genome. This is one of the most popular tasks in bioinformatics, and there many tools and pipelines available. You can use integrated suites like GALAXY or UGENE which contain everything you need for variant calling workflow, or compose your own pipeline using standalone tools.
In general, you should consider following key steps:
1. Quality control of sequenced reads (FastQC)
2. Read mapping on reference (BWA, Bowtie2)
3. Filtering and refinement of mapped reads (Picard suite)
4. Variant calling (bcftools, GATK, MrBayes)
5. Variant refinement (GATK)
6. Following variant processing and interpretation, according your needs.
Alexandr Pozharskiy Sir thank you, its a whole array of knowledge on a new dimention for me. Glad to know them.
Alexandr Pozharskiy Thank you so much with detailed explaination. That would be super helpful to me.
Xiyan Wang , you need Picard by the Broad Institute: https://broadinstitute.github.io/picard/
Here you can read about read processing involving Picard:
https://gatk.broadinstitute.org/hc/en-us/articles/360035535912-Data-pre-processing-for-variant-discovery
Alexandr Pozharskiy Thank you sooooo much! As you suggested, I have successed to detect the the mutation sites. One question about the mapping step: IIIumina raw data are alll short reads. If there is a large insertion sequence (1kb-2kb) that cant be mapped with reference. Will it be kicked out? If so, do you have any experiences how it could be detect?
Xiyan Wang , as I see, you mean if some of your strains have relatively long regions absent in the wild type genome. If so, yes, reads corresponding to this sequence will not be aligned. See Picard reference documentation to know how to obtain summary statistics of alignment.
If you suspect presence of such an insertion, you'd better look into existing literature on your species' genomic diversity, it could be described previously. If its position in the genome is known, it can easily be detected using PCR. Alternatively, you could try to BLAST or align your initially obtained contigs over the reference genome.
Alexandr Pozharskiy Thank you for your explaintions. I mean spontaneous mutation could due to the insertion sequences (it could a transponson). I have read some artcles that it could be happend on my bacteria. But they are not the same insertion sequences in the artcles, so that's why I would like to know if there are any insertion sequences in my spontaneous mutants. If I use de nove assembly for resequencing data, do you think I could use it to compare with wildtype? do you know any software could be used for this comparsion?
Xiyan Wang You should look for tools for multiple alignment, supporting very long sequences. However, it can very computationally intensive. Since I haven't worked directly with such long sequences as whole bacterial genomes or assembled contigs, I am not able to give you more precise instruction in this respect.