Region of zero coverage

[Sergioo]

Dear all,
I have been sequencing some samples and found something that is new for me.
After sequencing using Miseq, I usually align reads using CLC Genomic Workbench and from there, I get the visual of the mapping in general before proceeding to any downstream analysis.
Now, I've got some area of zero coverage and immediately though of sequencing errors. The problem is that this "hypothetical" error is seen in all my samples at a particular gene that is not seen by PCR in 1/2 of my samples (so it was reported as missing in PCR experimental data for those samples).
So, finally this let me think about just a kind of large deletion in the gene that has caused a significant change (for it not to be amplified by PCR) and altered presumably the function of of it. Does it make sense?
I think a gene may not be amplified in PCR for a number of reasons (no need to list them here), and obviously, sanger sequencing can not help. But for NGS, what should we expect? Please help brainstorming about this and see if I can come up with a new idea from the forum.
Please see attached the visualization of the mapping at that region (2 samples).
Thanks all,

[TiborNagy]

Yes, it can be a deletion.

[Sergioo]

Yes, it can be a deletion.

Thanks for your reply, but how do we know if it is a deletion or poor assembly/mapping ?

[GenoMax]

How big is this "deletion"? Are you seeing sharp edges at the boundaries where the reads stop being mapped to reference? Do these reads have parts that are left over?

You talked about attaching an image but I do not see it with your first post. Can you post that?

[Sergioo]

How big is this "deletion"? Are you seeing sharp edges at the boundaries where the reads stop being mapped to reference? Do these reads have parts that are left over?

You talked about attaching an image but I do not see it with your first post. Can you post that?

Thanks for your reply, Please see attached Pic 1 & 2., the deletion covers ~100,000 bp (The genome is ~3000 Mb) but in different area/regions. Please note that some regions correspond to genes that were not amplified using PCR (in a previous experiment) and thus the gene were considered missing.
Regards,

[GenoMax]

Please note that some regions correspond to genes that were not amplified using PCR (in a previous experiment) and thus the gene were considered missing.
Regards,

At least for these regions the NGS data should be additional evidence that they are indeed missing. (note: The thumbnails are hard to see so it is not clear if there are NO reads covering the deleted region).

Since these regions can be treated as positive controls for the rest of the data perhaps the other regions identified by NGS data are also missing (a hypothesis that would have to be proved by using PCR or something else).

[Brian Bushnell]

Thanks for your reply, but how do we know if it is a deletion or poor assembly/mapping ?

If you are interested in finding large deletions with mapping, I highly recommend you try BBMap. For a 3GB organism the command line would be like this:

(index)
bbmap.sh -Xmx24g ref=reference.fasta

(map)
bbmap.sh -Xmx24g in=reads.fq out=mapped.sam bs=bs.sh maxindel=200000 ambig=random

The "bs" flag is optional; it stands for "bamscript" and will tell BBMap to write a shellscript that, when executed, will use samtools to create a sorted indexed bam file which is needed by visualization programs like IGV.

Anyway, that command line will map and look for deletions of up to 200kbp. They will be contained within the reads' cigar strings (e.g. 40M100000D60M), so there will be no ambiguity about whether there is a deletion, or simply an area that happens to have no coverage.