---
tags: NCCUProject
---

# BlastFrost Figures

## Figure 2 - Comparisons of 4 methods for querying sequences
- BlastFrost(d = 1)
- at least 95% for alignments with at least 90% nucleotide identity
- high precision and sensitivity when genes length > 200 bp
- the least disk space for the genome indices of all these programs
- Megablast
- Megablast had low precision for genes length < 400 bp
- Blastn
- Both runtime and disk storage space are too high
- BIGSI(t = 0.4)
- If gene length < 200 bp, only BIGSI had high precision
- low sensitivity when nucleotide identities below 95%


---

## Figure 3 - ( Bifrost + BlastFrost ) VS Megablast
A. Megablast is faster in indexing step
B. BlastFrost is 10 times faster than Megablast in querying step
C. Bifrost Graph needs much less disk space
D. Megablast is faster for a single phase of creating an index 10,000 genomes plus a single round of querying up to 50,000 genes.

[ Summary ] Bifrost can rapidly and continuously expand its preconstructed graph to include additional genomes

## Figure 4 - BlastFrost VS BIGSI(Fig A,B), minimap2(Fig C,D)

A.B. BlastFrost(d=1) is faster than BIGSI search and used much less RAM for number of queries < 1200
C. BlastFrost is much faster than minimap2
D. The time needed for BlastFrost to extract a subgraph is dependent on the number of hits for that query


## Figure 5 - BlastFrost presence/absence analysis
Circle
- outermost circle: consists of one segment for each genome from Salmonella bongori and multiple subspecies of Salmonella enterica
- internal concentric circles: each of the genes within SPI-2

Color
- Dark green: Exact (d = 0)
- light green: Inexact (d = 2)

[ Summary ] Emphasized the importance of inexact querying, beacuse the inexact search increased the number of genes identified in the other subspecies.