Problem With Ucsc Liftover Tool

Question

[ancient_learner]

HI I was using UCSC liftover tool to lift the mm8 version to mm10. The following is the trial data that i gave. the difference between the start and end positions (2nd and 3rd columns) is around 1000bps.

chr17    14696791    14697791    
chr17    33422990    33423990        
chr17    27382883    27383883    
chr17    25971254    25972254        
chr17    44970114    44971114

However when i have converted to mm10 i observed something like this

**chr17    14960215    14992839**    
chr17    33817030    33818030    
chr17    27655294    27656294    
chr17    26243654    26244654    
chr17    45506533    45507533

The first row (highlighted with **) the difference between start and end was around 32 kb. Not sure what was happening. The options that i used were

Minimum ratio of bases that must remap: 0.95
Minimum hit size in query: 0
Minimum chain size in target: 0
Min ratio of alignment blocks or exons that must map: 1

and i kept Allow multiple output regions option unchecked

Answer 1

[Fred]

I tried the conversion of the first coordinates doing mm8 --> mm9 --> mm10

And for the first coordinates it gives:

1. mm8: chr17:14696791-14697791
2. mm9: chr17:15097220-15098220
3. mm10: chr17:14960215-14961215

Each times with a 1000bps difference.

Moreover, the DNA sequences (1) and (3) are the same (with ~6 mismatches)

[ancient_learner]

Thank you very much. Does this mean that i should not convert directly from mm8 to mm10? Because i have done many of my previous works this way will they be wrong?

[Vikas Bansal]

I think you should do it sequentially. Here is one answer I found from UCSC Genome Bioinformatics Group member.

That is exactly right: you should do sequential lifts from mm6 ->  mm8 ->  mm9 ->  mm10.

[Fred]

I don't know if it is generally better to liftOver coordinates with intermediate assemblies. May be someone else can give a better answer.