Bowtie2 sensitivity vs accuracy

[Surya Saha]

Hi Nicola,

I am working with a <70bp read data set with very low depth of coverage (4X - 18X). The default Bowtie2 settings of very-sensitive do not produce any hits to the marker database. However, I do get hits for very-sensitive-local. Do you have a sense of how the accuracy changes when more sensitive parameters like sensitive-local and very-sensitive-local are used? I may have missed it but I could not find a specific mention of this in the paper or supp data. Thanks much!

Best,

Surya

-----------------------------------------------------

Surya Saha
Department of Plant Pathology, 

Cornell University, NY, USA.

Sol Genomics Network, 

Boyce Thompson Institute, Ithaca, NY, USA.

http://www.linkedin.com/in/suryasaha

Twitter: @SahaSurya

[Nicola Segata]

Hi Surya,

 as a rule of thumb I'd say to stick with the global version of human-associated microbiomes for which the coverage of reference genomes is high, whereas for environmental microbiomes using the local version of the mapping could indeed produce better results. What kind of metagenomes are you looking at? I'm not sure what you mean with depth of coverage, are you maybe processing a single organism dataset?

many thanks

Nicola

[Surya Saha]

Hi Nicola,

This is a plant host-associated microbiome so it can be called a single organism data set. The depth of coverage is according to the host genome size. I am interested to see if I find any of the plant endophytes in the sequences we have. How does increasing the sensitivity of Bowtie2 to very-sensitive-local affect accuracy of MetaPhAln in your experience?

Please let me know if you have more questions. Thanks much!

Best,

Surya

--

You received this message because you are subscribed to the Google Groups "MetaPhlAn-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to metaphlan-use...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

[Nicola Segata]

Hi Surya,

 thanks for the clarification, I wrongly understood you were referring to a dataset with only one organism in the microbiome. In my experience, the non local version of BowTie2 is optimal for human-associated microbiomes whereas the local version (especially very-sensitive-local) is more accurate for environments, like plants or soil, in which it is less likely to have a high number of sequenced genomes close to the genomes of the organisms in the actual microbiome.

I hope this helps,

thanks

Nicola

[Surya Saha]

Hi Nicola,

Thanks for the input. This validates my experience too. Are you aware of any work/papers where MetaPhAln was used for plant or environmental samples?

Best,

Surya

-----------------------------------------------------

Surya Saha
Department of Plant Pathology, 

Cornell University, NY, USA.

Sol Genomics Network, 

Boyce Thompson Institute, Ithaca, NY, USA.

http://www.linkedin.com/in/suryasaha

Twitter: @SahaSurya

[Nicola Segata]

Hi Surya, no, I'm not aware of MetaPhlAn applications on plant-associated microbiomes, but I'm not familiar with the literature in that field...

thanks

Nicola

[Surya Saha]

I will look around. Thanks for all the help, Nicola.

Best,

Surya