[Inparanoid] confidence value cutoff

Pengcheng Yang yangpc at biols.ac.cn
Tue Sep 29 01:51:19 CEST 2015


Hi Erik Sonnhammer,

Thank you for your reply.
One phenomenon is found that for the two organisms A and B with A-1 and 
B-1, B-2 proteins, respectively. These three proteins clustered into one 
ortholog group.
I found that the parsed blast results contain the hit records between 
A-1 and B-1, B-1 and B-2, but not A-1 and B-2. However, this is possible 
because blast_parser.pl set several criteria to filtering the alignment.

Ok, I will leave it at 0.05.

Thank you.

Pengcheng

On 2015/9/29 0:46, Erik Sonnhammer wrote:
> Hi,
>
> 0.05 is "optimised for coverage". We have not actually done any proper 
> optimisation of it, but we know that the false positive rate is low at 
> this value, so one could even try to lower it.  But if you only want 
> the very strongest orthologs then you can increase it.  On 
> http://orthology.benchmarkservice.org we use default cutoff and 1.0.  
> The main effect is that 1.0 gives much lower coverage.  Sometimes 1.0 
> has better accuracy but sometimes not.  So unless you are particularly 
> interested in 1-1 orthologs I recommend leaving it at 0.05.
>
> Erik Sonnhammer
>
> On 2015-09-28 16:42, Pengcheng Yang wrote:
>> Hi Inparanoid developers,
>>
>> I am using Inparanoid to cluster proteins into groups to identify 
>> ortholog genes. However, I have a questions about the confidence 
>> value cutoff. Is it too low, default 0.05 to increase the false 
>> positive? What is the optimized value you recommend?
>>
>> Thank you and best wishes,
>>
>
> _______________________________________________
> InParanoid mailing list
> InParanoid at lists.su.se
> https://lists.su.se/mailman/listinfo/inparanoid-at-sbc.su.se




More information about the InParanoid mailing list