LSA_output; several LS for same OTU pairs

[Former user]

Hi, I have some issues to understand and analysis the output of an LSA analysis. I have run 'lsa_compute ' via 'par_ana' on a sequencing dataset as follow:

python ~/charade-elsa-80c7298487ce/lsa/par_ana.py Raw_abundance_OTUs_dom.txt Raw_abundance_OTUs_dom.lsa 'lsa_compute.py %s %s -e Raw_abundance_OTUs_dom.txt -r 1 -d 0 -s 7 -b 0 -n robustZ -p theo' $PWD

Here, I would like to analysis the OTUs cooccurrence in 7 sites (delay =0, then).

After filtering of the output files (based on p and q values), I haver noticed that for some OTU pairs, I have several LS scores. For instance, i have:

X Y LS lowCI upCI Xs Ys Len Delay P PCC Ppcc SPCC Pspcc Dspcc SCC Pscc SSCC Psscc Dsscc Q Qpcc Qspcc Qscc Qsscc Xi Yi 
OTU_1380    OTU_12316   -1.083529   -1.083529   -1.083529   6   6   1   0   0.016419    -0.128134   0.80885 -0.128134   0.80885 0   -0.054772   0.917924    -0.054772   0.917924    0   0.005144    0.999432    0.999432    0.002234    0.002234    1   2826
OTU_1380    OTU_12316   1.069152    1.069152    1.069152    6   6   2   0   0.018619    0.620174    0.189003    0.620174    0.189003    0   0.602495    0.205611    0.602495    0.205611    0   0.005144    0.982799    0.982799    0.000564    0.000564    1   989
OTU_1380    OTU_12316   1.069152    1.069152    1.069152    6   6   2   0   0.018619    0.620174    0.189003    0.620174    0.189003    0   0.602495    0.205611    0.602495    0.205611    0   0.005144    0.982799    0.982799    0.000564    0.000564    1   98
OTU_1380    OTU_12316   1.069152    1.069152    1.069152    6   6   2   0   0.018619    0.620174    0.189003    0.620174    0.189003    0   0.602495    0.205611    0.602495    0.205611    0   0.005144    0.982799    0.982799    0.000564    0.000564    1   929
OTU_1380    OTU_12316   1.069152    1.069152    1.069152    6   6   2   0   0.018619    0.620174    0.189003    0.620174    0.189003    0   0.602495    0.205611    0.602495    0.205611    0   0.005144    0.982799    0.982799    0.000564    0.000564    1   920
OTU_1380    OTU_12316   1.069152    1.069152    1.069152    6   6   2   0   0.018619    0.620174    0.189003    0.620174    0.189003    0   0.602495    0.205611    0.602495    0.205611    0   0.005144    0.982799    0.982799    0.000564    0.000564    1   695
OTU_1380    OTU_12316   1.069152    1.069152    1.069152    6   6   2   0   0.018619    0.620174    0.189003    0.620174    0.189003    0   0.602495    0.205611    0.602495    0.205611    0   0.005144    0.982799    0.982799    0.000564    0.000564    1   664
OTU_1380    OTU_12316   1.069152    1.069152    1.069152    6   6   2   0   0.018619    0.620174    0.189003    0.620174    0.189003    0   0.602495    0.205611    0.602495    0.205611    0   0.005144    0.982799    0.982799    0.000564    0.000564    1   431

How should I interpret such results? In this example, all the results are ordered according to their p and q values. Here, the first line (best p and q values) has a negative LS score while the other ones are all positive. It is quite confusing. Best,

[Charlie Xia]

Issue #9 was marked as a duplicate of this issue.

[Charlie Xia]

Hi. Thanks for raising the issue. Can you attach a minimal input to replicate the error. On the meanwhile, you can filter out the duplicated lines if they are just identical to proceed your project.

Thank you.

[jeff63]

Hi, And thank you for your reply. Here is the first lines of my input dataset:

#OTU_ID t1      t2      t3      t4      t5      t6      t7
OTU_12283       40      8       935     3       157     80      345
OTU_9775        390     76      11      86      487     513     2
OTU_751 4       38      33      18      49      3       1419
OTU_11718       0       8       0       0       1       0       1554
OTU_6787        12      83      607     523     3       170     158

Indeed, it seems that most of the lines are duplicated (solely the last output criteria (Yi) may varied between them). My question was concerning the case when I may have two different output results for the same OTUs pairs. I mean "different" when the LS scores can be either negative or positive for a relative similar p/q values. I have solely this "case" 15 times for a total of around 20000 edges but I wonder what values should I trust (which may change the biological meaning of such OTUs pairs). In each case, I have only one line with a negative LS score for a lot of duplicated lines with positive and identical LS score.

Thanks again for your help.

[Charlie Xia]

lsa_compute ../test/multiLS.txt ../test/multiLS.lsa -r 1 -s 7 -d 3 -p theo -x 1000 -f none -n percentileZ -e ../test/multiLS.txt -m 0

tried your file with above command, cannot replicate error. Mark as invalid and close.