Thank you very much. I will surely try this method.
• 0 views
•
link
written
by
thakurshalabh •
Clustering Gene Absence And Presence Data In Binary Format
Hi, I have gene absence and presence data for approximately 60 genomes. I have created matrix for each gene family by giving value 1 if its present and 0 if its absent. I want to cluster this data by strains which are more similar in sharing genes and also gene families which are shared in different strains. I know R can do Hierarchical clustering. But I am looking for some thing more visual such as heat map or correlation plot.
My data look like this. Any idea what method would be best to represent this data?
GROUP Pla302278PT Pla3988 PmaH7608 Pma90_32 PtoDC3000 PmaM6 PmaM4a Pto1108 PtoT1 PtoK40 PtoMax13 Pav631 Pmp302280PT Pan302091 Ptt50252 Pja301072 Ppi1704B Pav037 Pav013 Pac302273 PacA10853 PsyB728a PssB48 PssA2 PsCit7 Psv4352 Psv3335 PmyAZ84488 PmpFTRS_U7 Pae3681 Pae0893_23 Pla301315 Pla107 PlaYM7902 Pta11528 Pta6606 PseHC_1 Pmo301020 PgyB076 PgyUnB647 PgyBR1 PgyKN44 PgyLN10 PphY5_2 PmaKN91 PphNPS3121 PphHB10Y Pph1448A PmeN6801 Pph1302A PmaYM7930 PmaES4326 Pci0788_9 Por36_1
OrthoGroup7591.1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 3 3 3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
OrthoGroup13947.1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 3 0 0
OrthoGroup6352.114 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 -1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
OrthoGroup3637.2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
• 8,760 views
•
link
updated
by
Naren
•
written
by
thakurshalabh •
•
written
by
thakurshalabh •
3 answers
You can solve this problem in a very easy method,
I explained very briefly here, This can be useful for your analysis. You will get lot of material on various methods and algorithms in R. No need to code yourself
Step-1: Construct a formula, using which you can calculate distance(using presence of absence of data) between the genome of various strains. Calculate genome distances pair-wise.
Step-2: Subject the Distance data to PCA analysis.
Step-3: From PCA analysis output find the minimum number of dimensions which explains variablity upto atleast 80%(May be first 15-20 dimensions).
Step-4: Using any unsupervised clustering like k-means to cluster(No. of clusters should be preset, you can tune the parameters accordingly) them.
Step-5: Take cluster centers construct a phylogenetic tree.
Now you will have the releationships among the different genome(belonging to different clusters). Hope this helps
• 110 views
•
link
written
by
aravind ramesh
All the best. I am curious to see your results. If possible post the final tree here.
• 0 views
•
link
written
by
aravind ramesh
The R heatmap function might be of use to you. Also, you might try this search:
• 0 views
•
link
written
by
Sean Davis
1.If you are on windows, Download Past.exe
2.import data file in tab delimited text format.
3.Go to 'Multivar' menu.
4.Choose 'Cluster analysis'.
5.save cluster in nexus format or copy graphic to an image.
(Your data seems to contain 333 and -1 such numbers are confusing, it should be in 101010 format for correct clusters)
Enjoy.
Log in to answer this question.
Please look carefully at your data and make sure what is shown above is the correct format (I was assuming your data is not in paragraph form). Also, just post a sample, not the whole data set.