Understanding Numbat results

[acalcino]

I'm struggling a bit to understand what final output file I should consider for downstream analyses following a numbat run. I ran numbat with three iterations and I now want to extract all the cell barcodes associated with a particular genotype. If I run this:

cut -f3 ../clone_post_3.tsv | sort -u

I can see there are six genotypes:

""
13a,4c_bamp,18a
13a,4c_bamp,18a,15b,2a
13a,4c_bamp,18a,15b,2a,10a,4a,5a,3a
13a,4c_bamp,18a,15b,2a,10a,4a,5a,3a,6a,7a
13a,4c_bamp,18a,15b,2a,1c,4c_loh

I'm particularly interested in the 4c_loh (last genotype) so when I extract all the barcodes associated with this:

cut -f1,3 ../clone_post_3.tsv | grep -P '13a,4c_bamp,18a,15b,2a,1c,4c_loh$' | wc -l

I get 1555 barcodes which is consistent with the numbers on the clone_post_final.png figure.

If I try to do the same thing from within R:

nb = Numbat$new(glue('/g/data/pq84/single_cell/10X/2046t2/numbat/trial_1/'))
nb$clone_post %>% distinct(GT_opt)

GT_opt
1:
2: 18a,13a,4d_bamp,2a
3: 18a,13a,4d_bamp
4: 18a,13a,4d_bamp,2a,10a,3a,5a
5: 18a,13a,4d_bamp,1c,4d_loh
6: 18a,13a,4d_bamp,2a,10a,3a,5a,6a
7: 18a

nb$clone_post %>% filter(GT_opt == "18a,13a,4d_bamp,1c,4d_loh") %>% select(cell, GT_opt)

I get 1266 barcodes as opposed to the 1555 from the clone_post_3.tsv file. Also, the 4c_bamp from clone_post_3.tsv is called 4d_bamp here and there are seven genotypes, not six.

So my question is, which one is correct? Should I be extracting barcodes using the Numbat object in R or should I be extracting them from the clone_post_3.tsv file? Why are there differences between these two?

[teng-gao]

Hi @acalcino ,

There's an i parameter in Numbat$new that determines which iteration's results to use when creating the Numbat object (default is 2). In your case should be i=3.
https://kharchenkolab.github.io/numbat/reference/Numbat.html#method-Numbat-new

[acalcino]

Thanks very much, I hadn't realised it was defaulting to i=2. Much appreciated!