Khalid Usman

E.g. if your original input matrix has 1,000,000 number of cells and 100 genes. You don't want to process all rows, so you can perform either uniform sampling or weighted...

Hi Philipp, I have updated accordingly, again no issue but the duplication and i analysed most of them are from previous code. Regards, Khalid On Mon, May 20, 2019 at...

1. No, I have sampled cells with weights, out of those 1000 rows most having weight=1, e.g. 1 row has weight 125, then in gene ranking the expression all genes...

I am using a sampling technique, which samples few rows without descreasing performance. So speed is more than 10X time faster for larger dataset with similar accuracy. On Tue, May...

Yes , the sampling is done with weights and I used the coreset technique for it. On Tue, May 21, 2019 at 5:29 PM MalteDLuecken wrote: > I understand the...

Exactly, It's similar as Philipp explained. Regards, Khalid On Tue, May 21, 2019 at 8:31 PM Philipp A. wrote: > Not clusters, I meant that cells are selected to represent...

So, what's next ? :) I think no more issues in code , but little duplications. Thanks On Tue, May 21, 2019 at 10:06 PM khalid usman wrote: > Exactly,...

Thanks , But i will suggest to just support weights instead of coreset, may be user want to sample data with some other weighting technique. So we should ask them...

Thanks Philipp, I have updated and push the code. I hope you will accept this pull request now. To support PCA and scanpy for weighted sampling, you can just set...