Table 3 Loci included in the MD-CNN and SD-CNN models.

From: A convolutional neural network highlights mutations relevant to antimicrobial resistance in Mycobacterium tuberculosis

Locus

Start

End

Drug(s)

Length (in H37Rv)

acpM-kasA

2,517,695

2,519,365

Isoniazid

1670

gid

4,407,528

4,408,334

Streptomycin

806

rpsA

1,833,378

1,834,987

Pyrazinamide

1609

clpC

4,036,731

4,040,937

Pyrazinamide

4206

embCAB

4,239,663

4,249,810

Ethambutol

10,147

aftB-ubiA

4,266,953

4,269,833

Ethambutol

2880

rrs-rrl

1,471,576

1,477,013

Streptomycin, Amikacin, Capreomycin, Kanamycin

5437

ethAR

4,326,004

4,328,199

Ethionamide

2195

oxyR-ahpC

2,725,477

2,726,780

Isoniazid

1303

tlyA

1,917,755

1,918,746

Capreomycin

991

katG

2,153,235

2,156,706

Isoniazid

3471

rpsL

781,311

781,934

Streptomycin

623

rpoBC

759,609

767,320

Rifampicin

7711

fabG1-inhA

1,672,457

1,675,011

Isoniazid, Ethionamide

2554

eis

2,713,783

2,716,314

Kanamycin, Amikacin

2531

gyrBA

4997

9818

Ciprofloxacin, Levofloxacin, Moxifloxacin, Ofloxacin

4821

panD

4,043,041

4,045,210

Pyrazinamide

2169

pncA

2,287,883

2,289,599

Pyrazinamide

1716

  1. The 18 loci included in the MD-CNN and their start and end coordinates (in H37Rv numbering). Each locus is designated as putatively involved in resistance to at least one drug. To construct the 13 SD-CNN models, the relevant loci for each drug are combined—for example, the isoniazid (INH) model contains the acpM-kasA, oxyR-ahpC, katG, and fabG1-inhA loci.
Back to article page