PD_whole_tree vs Chao1 vs Observed species

[Iswan]

When I done the ANOVA (data pull out from alpha.div text), each parameter of PD_whole_tree, Chao1 and Observed species give different signifcant value. I also found by glance observation, it corespond with the rarefiction curve (attached). 

In PD_whole _tree, all group didn't show any significant difference.

In Observed Species, both Cam_Co_C and Cat_Co_C significantly higher (p<0.05) than Cam_Alone_C.

In Chao1, each group were significantly difference(p<0.05) to each other by Cat_Co_C the highest, follow by Cam_Co_C the middle and the lowest one by Cam_Alone_C. 

Does this happen because the species richness (measured by Chao1) is more depth than the count of unique OTUs (measured by Observed species). 

Is there a treshold difference for each parameter? Anyone can suggest a good source for better understanding between diversity (PD_whole_tree), species richness (Chao1) and Observed Species (OTUs)

Thanks.. 

[Iswan]

Sorry for the typo. 

Cam_Co_M (in the pic attached) = Cam_Co_C

Cam_Alone_M (in the pic attached) = Cam_Alone_C

Cat_Co_M (in the pic attached) = Cat_Co_C

[Luke Ursell]

Hey Iswan,

One issue you have at the moment is that your rarefaction curves are not covering the same sequences per sample - you can see how your blue line has far more sequences per sample than the others lines. So I would suggest you redo your alpha_rarefaction.py and pass in the -e for max rarefaction depth. Looking at your curves I might pass in -e 1400

The other thing is that QIIME has a script called compare_alpha_diversity.py that can tell you if one of the curves has a significantly higher alpha diversity at a given rarefaction depth. I fear that you might be getting spurious results if you are importing all of the alpha rarefaction numbers into another program and running ANOVA because you have not equally sampled all of your sequences at the same depth.

As far as understanding the metrics, I recommend you search the forum for other posts explaining the difference between the three, or look up some of the original publications.

Hope this helps

Luke

--
 
---
You received this message because you are subscribed to the Google Groups "Qiime Forum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to qiime-forum...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
 
 
<Observed species Meramist.png><Chao1 Meramist.png><PD Meramist.png>

[Sergio Balzano]

Hello everybody

I have a question similar to that reported by Iswan some time ago. My rarefaction curves show contradictory results according to the parameter used and I was wondering if anyone had similar results in his/her data or if you have any suggestion in interpreting my data.   

Please find attached the graph including all rarefaction. Samples Although there are a couple of samples which show systematically a high diversity and a low diversity, this is not the case for all. By the way chao and OTU curves are very similar because I removed singleton from my dataset  prior to analyses (and as far as I have understood if there are no singletons chao1 index is the same as number of OTUs).

1) My interpretation is that sample Coo1Jun has an enormous number of OTUs compared to the other samples, but not such a high diversity.

   Do the curves suggest that most OTUs found in Coo1Jun are very poorly abundant, whereas in contrast sample Coo1Nov has fewer OTUs most of them without very low relative abundances?

2) For what I read PD_whole_tree corresponds to the sum of all horizontal distances from the root, for all the sequences in a given sample, therefore this parameter does not include the relative proportion of each OTU, is that right? in that case sample Coo5Nov would be the most phylogenetically diverse in my dataset, although it doesn't have the highest number of OTUs neither the highest Shannon/simpson indices.

Thanks for your attention

regards

Sergio

[Will Van Treuren]

Hi Sergio,

> 1) My interpretation is that sample Coo1Jun has an enormous number of OTUs compared to the other samples, but not such a high diversity.

   Do the curves suggest that most OTUs found in Coo1Jun are very poorly abundant, whereas in contrast sample Coo1Nov has fewer OTUs most of them without very low relative abundances?

I think that your interpretation is correct. There are a large number of low abundance OTUs in the Coo1Jun sample. Coo1Nov has fewer, more evenly abundant (between them) OTUs. 

2) For what I read PD_whole_tree corresponds to the sum of all horizontal distances from the root, for all the sequences in a given sample, therefore this parameter does not include the relative proportion of each OTU, is that right? in that case sample Coo5Nov would be the most phylogenetically diverse in my dataset, although it doesn't have the highest number of OTUs neither the highest Shannon/simpson indices.

Correct I think. PD_whole_tree measures the total descending branch length for the OTUs in your sample. Coo5Nov appears to have a large amount of phylogenetic diversity, even though it has a relatively smaller number of OTUs. This suggests to me that it probably is a much more diverse sample (i.e. it has OTUs from many different clades in the tree) as opposed to Coo1Jun which has a lot of low abundance OTUs from only one or two clades in the tree. 

Hope that helps,

Will 

--

---
You received this message because you are subscribed to the Google Groups "Qiime Forum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to qiime-forum...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Sergio Balzano]

Hi Will

thanks for your suggestion, all these explanations make sense to me

best regards

Sergio