BWA Cleaner/Fixer/Filter to get rid of "spikes"

Has anyone thought about taking the ensemble average of multiple BWA data to produce a “clean” data set free of spikes?

I could easily write a program to take, say, four BWA files, averaging among the four to get rid of spikes. Or instead of averaging, I could drop the data point that is not like the other three.

So if any of you know the file format for BWA files, please let me know so I can try this out.


Someone wrote a program which lets you edit the BWA to manually get rid of the spikes. Its available here.

But please, dont let that deter you from making your program, it will definatly be good :slight_smile:

That would be great to see a BWA fixer. But I’d imagine it to be something that will take just one halfway crappy BWA and fix it.

It would basically follow this scheme. Base curve being the main line and doublesectors being the small raised squiggles.

1 Find the base curve of the BWA (calculate what the curve should be from basic data in the BWA)

2 Reform that curve so that no section are outside of the max and min (get ride of all spikes outside the small range of a normal BWA’s base curve and doublesectors sections)

3 Distinguish sections that are part of the base curve from those that are doublesectors and rebuild them (find the mostly high squiggly sections, assume they are double sectors if their length is ‘typical’ and restore their integrety / amplitute)

So basically it would be pretty easy to get rid of the major spikes once you’ve determined the base curve. The hard part is measuring the length of the doublesectors section and restoring them. Everything would have to take example from all the perfect BWA’s that have been built already.

This is just the stuff that comes out of my butt once and awhile.

Originally posted by Sunork
That would be great to see a BWA fixer. But I’d imagine it to be something that will take just one halfway crappy BWA and fix it.

There is nothing wrong with that! I recommend to anyone to use that BWA edit program on seemingly perfect BWA files too. Tidy up any slight spikes and the access time prior to the game loading is significantly reduced.

The home page of BWA Edit is here: (url] The server is way too slow but it works.

@calpchen: After many tests, with some CDs, the spikes appear always at the same place: there’s no way to correct them with a multiple BWA extraction. The best way should be to seek & destroy them by scanning the data (there’s no possible confusion between a spike and sector density variations). But I’ve no time to implement it yet.

I’m happy to see that some guys are using BWA Edit. Id’ like some feedback about it. Anyway, the source code is provided with BWA Edit so feel free to enhance it yourself if you wish !

BTW I’m looking for a free hosting with no banners (free as free beer and free speech ;)), and quicker than *