Could i use gfase for Flye (HERRO corrected ul-ONT)

[YouxinZhao]

Could i use gfase for gfa of Flye (HERRO corrected ul-ONT reads)?

[rlorigro]

Hi, apologies for the late reply. The answer is likely no, because in the past Flye did not generate a GFA in the late stages of assembly that would have sufficiently resolved haplotypes for mapping reads. However, if that has changed, then I would refer to these requirements from the publication:

2. The cumulative consumed query/target length of a
   GFA overlap cigar must not exceed the length of the
   query/target node. This is not possible in a properly format-
   ted GFA, but it has been observed in both hifiasm and Verkko
   overlaps. If this input error is encountered, it is recommended
   to use the ‐‐skip_unzip argument.
3. Edges (L lines) are not required. For phasing an input that
   is not in graph form (e.g., in FASTA/Q format), it is possible to
   simply convert it to a GFA without any L lines, and invoke
   the ‐‐skip_unzip and ‐‐use_homology flags to bin the contigs
   into bin 1 or 2 or unphased. Homology will be inferred entirely
   independently of the graph edges.
4. When using an alignment as input, nodes (“segments”
   or S lines) in the graph should be mappable. Short nodes
   that are insufficient in length to map a Hi-C or Pore-C subread
   (≥150 bp of nonrepetitive sequence) will not accumulate any
   contacts in the mapping step. Nodes without contacts will
   not be phased, and GFAse will put them in the “unphased”
   bin instead of using them to extend a phased chain. In addition
   to this, when using the homology-based alt detection, nodes
   must be long enough to confidently map to each other. We rec-
   ommend starting with an assembly configuration that produc-
   es a phase N50 as large as possible without introducing any
   switch errors. For GFAs in which nodes are not mappable, the
   user may provide a custom contact map CSV as input.