Checking for Mycoplasma contamination

[id0]

I am processing some sequencing data that came from cell culture samples. There is some potential Mycoplasma contamination. If you are not familiar with working with cells, this is a very common occurrence. I think the easiest solution would be to align to Mycoplasma and see how many reads actually align. Unfortunately, there are many Mycoplasma species and genomes seem to vary widely. Is there already a pre-compiled list of reference sequences I can align to in order to check for that? Are there better solutions to check for contamination?

[rhinoceros]

How's the GC distribution of your reads?

[id0]

How's the GC distribution of your reads?

It's normal, which is about 40%.

[winsettz]

I am processing some sequencing data that came from cell culture samples. There is some potential Mycoplasma contamination. If you are not familiar with working with cells, this is a very common occurrence. I think the easiest solution would be to align to Mycoplasma and see how many reads actually align. Unfortunately, there are many Mycoplasma species and genomes seem to vary widely. Is there already a pre-compiled list of reference sequences I can align to in order to check for that? Are there better solutions to check for contamination?

If I remember correctly, in the lab the way to do it is by PCR. In silico, you could recapitulate that by finding the gene that is PCR'ed and screen your reads for it. Can't say what the FP/FN rate would be, but it'd be a start.

[id0]

If I remember correctly, in the lab the way to do it is by PCR. In silico, you could recapitulate that by finding the gene that is PCR'ed and screen your reads for it. Can't say what the FP/FN rate would be, but it'd be a start.

Then you are left with an additional step of trying to quantify the amount of Mycoplasma.

[winsettz]

Then you are left with an additional step of trying to quantify the amount of Mycoplasma.

I suppose if you had the original samples, the experimental way to do it would be quantitative PCR. With reads, you could probably determine nucleotide coverage of your Mycoplasma marker, or ideally, multiple mycoplasma markers, and compare to coverage of a marker of your desired organism.

[simonandrews]

You can get a list of all of the available mycoplasma genomes at:

http://www.ncbi.nlm.nih.gov/genome/?term=Mycoplasma

You could combine these into a single search database and just look for hits to any parts of any of them in your screening. I don't know how complete these sequences would be in covering the diversity of mycoplasma species but if you included them all and allowed some flexibility in matching that would hopefully be fairly useful?

To do the searches we developed FastQ Screen which allows you to search through multiple databases and provides a combined report on the relative number of unique and multiple hits to each of them. It's an easy way to get quantitative or visual feedback of the amount of contamination in your libraries.

[id0]

To do the searches we developed FastQ Screen which allows you to search through multiple databases and provides a combined report on the relative number of unique and multiple hits to each of them. It's an easy way to get quantitative or visual feedback of the amount of contamination in your libraries.

I was not aware of that tool. It looks very promising. It seems it should be a part of every sequencing pipeline.