I have a problem, I need to parse the following dataframe:
cluster_name qseqid sseqid pident_x qstart qend sstar send
2 1 seq1_0035_0035 seq13_0042_0035 0.73 42 133 46 189
3 1 seq1_0035_0035 seq13_0042_0035 0.73 146 283 287 389
4 1 seq1_0035_0035 seq13_0042_0035 0.73 301 478 402 503
5 1 seq13_0042_0035 seq1_0035_0035 0.73 46 189 42 133
6 1 seq13_0042_0035 seq1_0035_0035 0.73 287 389 146 283
7 1 seq13_0042_0035 seq1_0035_0035 0.73 402 503 301 478
8 2 seq4_0042_0035 seq2_0035_0035 0.71 256 789 125 678
9 2 seq4_0042_0035 seq2_0035_0035 0.71 802 1056 706 985
10 2 seq4_0042_0035 seq7_0035_0042 0.83 123 745 156 723
12 4 seq11_0035_0035 seq14_0042_0035 0.89 145 647 236 921
13 4 seq11_0035_0035 seq17_0042_0042 0.97 148 623 241 1002
14 5 seq17_0035_0042 seq17_0042_0042 0.94 188 643 179 746
and only get within each cluster the maximum pident_x but the issue is that as you can see I can have reversed sequences (if you take a look at the 2,3,4 and 5,6,7 they are the same but reversed) and what I need to do is to keep only one for exemple only the line 2,3 and 4.
The output would be then :
cluster_name qseqid sseqid pident_x qstart qend sstar send
2 1 seq1_0035_0035 seq13_0042_0035 0.73 42 133 46 189
3 1 seq1_0035_0035 seq13_0042_0035 0.73 146 283 287 389
4 1 seq1_0035_0035 seq13_0042_0035 0.73 301 478 402 503
10 2 seq4_0042_0035 seq7_0035_0042 0.83 123 745 156 723
13 4 seq11_0035_0035 seq17_0042_0042 0.97 148 623 241 1002
14 5 seq17_0035_0042 seq17_0042_0042 0.94 188 643 179 746
Indeed :
for the cluster1:
seq1_0035_0035 vs seq13_0042_0035 has his reversed seq13_0042_0035 seq1_0035_0035 but I only keep the first one.
for the cluster2:
seq4_0042_0035 vs seq7_0035_0042 (0.83) has a better pident score than seq4_0042_0035 vs seq2_0035_0035 (0.71)
for the cluster4:
seq11_0035_0035 vs seq17_0042_0042 (0.97) has a better pident score than seq11_0035_0035 vs seq14_0042_0035 (0.89)
for the custer5:
There is only one paired sequence seq17_0035_0042 vs seq17_0042_0042 (0.94) , then I keep this one
I do not really know how to manage to do such a thing, someone has an idea?
•
written
by
Chvatil
1 answer
- add a new column in the dataframe
- for each row, the value in the new column is the set contains values of qseqid and sseqid. for example, the value of the new column in the first row is {seq1_0035_0035, seq13_0042_0035}.
- drop duplicates using subset: cluster & new column
- remove the new column
- do whatever you want
Log in to answer this question.
Your reversed co-ordinates only apply to cluster-1. Is there a different criterion you're using for clusters 2 and 4 in addition to simply picking the
max(pindel_x)value rows?Hi, the reversed co-ordinates apply on all cluster and In addition to keep only the best pident I also take care to have only sequences comming from different species so with the first _number different between two species.
You're going to need to explain the formats better. "The first _ number" is not really clear. Also, how do you see coordinate reversal in cluster-2?
Side note: This is purely a programming question - you seem to know your requirements and just need help implementing them, I'd say Google is your best friend because by figuring out the best search terms to find your solution, you'll know how to solve problems quicker, and if you were to depend on people, it'd be like asking for handholding.
I've not used pandas, but I can definitely help you out because this looks like a simple data-structure manipulation problem.