# OLD F. cylindrus assembly

# Initial graph and workspace

Graph at k=63, tipclipped at 125bp x3. Both GFAs dumped:

%cd /Users/clavijob/fcyl_pe2/sdgdbg/
import SDGpython as SDG
from graphcleaning import *

K=63
MIN_CVG=2
NUM_BATCHES=8

ws=SDG.WorkSpace()
peds=ws.add_paired_reads_datastore('fcyl_pe2.prseq')
SDG.GraphMaker(ws.sdg).new_graph_from_paired_datastore(peds,K,MIN_CVG,NUM_BATCHES)
ws.sdg.write_to_gfa1("fcyl_pe2_k63c2.gfa")
simple_structures_stats(ws.sdg)
c=SDG.GraphContigger(ws)
c.clip_tips(125,3)
ws.sdg.write_to_gfa1("fcyl_pe2_k63c2_tc125x3.gfa")
simple_structures_stats(ws.sdg)

Graph reloaded, KCI and paired read mapping:

%cd /Users/clavijob/fcyl_pe2/sdgdbg/
import SDGpython as SDG
from graphcleaning import *
from graphuntangling import *
from graphstriding import *

ws=SDG.WorkSpace()
ws.sdg.load_from_gfa("fcyl_pe2_k63c2_tc125x3.gfa")
ws.sdg.write_to_gfa1("fcyl_pe2_initial.gfa")

peds=ws.add_paired_reads_datastore('fcyl_pe2.prseq')
lords=ws.add_long_reads_datastore('fcyl_nano.loseq')

kc=ws.add_kmer_counter("main")
kc.add_count("pe",peds)
kc.set_kci_peak(61)
kc.compute_all_kcis()
print(ws.sdg.stats_by_kci())
ws.dump("fcyl_pe2_initial.sdgws")

peds.mapper.path_reads()
peds.mapper.dump_readpaths("fcyl_pe2_initial_readpaths.dump")

| KCI | Total bp | Nodes | Tips | N25 | N50 | N75 |
|-------|--------------:|---------:|--------:|-----------:|-----------:|-----------:|
| None | 30272273 | 407162 | 9961 | 87 | 72 | 64 |
| < 0.5 | 29221328 | 210177 | 9995 | 125 | 125 | 125 |
| ~ 1 | 41232715 | 175140 | 1354 | 552 | 257 | 149 |
| ~ 2 | 35931827 | 169937 | 367 | 445 | 220 | 127 |
| ~ 3 | 16008585 | 82284 | 52 | 396 | 191 | 125 |
| > 3.5 | 2554306 | 15334 | 89 | 276 | 147 | 125 |
| **All** | 155221034 | 1060034 | 21818 | 318 | 127 | 124 |

## Unused read coverage (idea).

Use 63-mers to start with. Only 63-mers of "unused" reads contribute to the coverage of a 63-mer (either on a KC or over the nodes, I am not sure). Once a read is "consumed" (i.e. it is fully assigned to a node), all the coverage of its constituen 63-mers is removed from the problem. Alternatively, a new DBG could be constructed by using only reads that are unused so far.