ram-compbio / cando Goto Github PK

• Index out of bounds
• variable n to topX for consistency
• add -1 as default for output

Create functions to calculate common metrics based on outcomes of canbenchmark_new (particularly NDCG, recall, precision, AUROC if possible)

Update README.md with the option to download the anaconda env for CANDO (all packages installed) and a preconfigured CANDO env with all packages and the newest stable version of CANDO.

e.g., cmpd_set in canpredict should be cmpd_lib, or something similar.

An artifact of v1 is lack of header for the matrix file. We need to add a header with either CANDO ids or compound names to enable more flexible use of the matrix without losing track of compound data.

• Add headers to matrices
• Add functionality to BANDOCK for column names/ids
• Modify CANDO init to handle the column name when creating Compound objects

So get_data() works correctly

Kindly help to solve the following error:
Generating CANDO matrix...
Traceback (most recent call last):

File "/tmp/ipykernel_3652/1890496437.py", line 10, in
lig_name=False, ncpus=ncpus)

File "/home/mi/anaconda3/lib/python3.7/site-packages/cando/cando.py", line 4224, in generate_matrix
p_dict[p[0]] = list(zip(p[1].split(','),p[2].split(',')))

IndexError: tuple index out of range

I would like to ostensibly replace the flat files with these portable dbs.

This will be an additional variable that will be present in the mapping file and will be processed on init. Each compound will have an associated source (public db, design protocol, etc.) from which it was extracted.

• Add "source" column to compound mapping file
• Add new variable to Compound class that will be populated by "source" during CANDO init
• Optional: Output "source" in canpredict and other functions

canbenchmark_ddi and canpredict

Adding "return" statements to a couple key functions (particularly canpredict) to make them more compatible with external functions, so you don't need to write to file and then read from file to get results if you're not working within the CANDO object

Created a new benchmarking function called canbenchmark_new() which uses a methodology similar to canpredict_compounds() to assess performance using the classic CANDO metrics (i.e., AIA, APA, and IC), which all are prepended with n to denote the new benchmarking protocol (e.g., nAIA, nAPA, nIC).

Add parameter to calculate additional metrics from the functions above as part of canbenchmark_new

Removed the references to objects within objects to reduce memory usage and depth.
Check all functions to confirm they work properly.

UnpicklingError: pickle data was truncated

Dear Sir,
I want to run the following code:
(https://github.com/lamawmouk/Tox21_FeatureSelection_SMOTEENN_RF/tree/main/Featuregeneration_using_CANDO)/Feature_generation_trainset.ipynb
.............................
import csv
with open('train_smile_toxnum.csv','r') as csvin, open('train_smile.tsv', 'w') as tsvout:
csvin = csv.reader(csvin)
tsvout = csv.writer(tsvout, delimiter='\t')

for row in csvin:
    tsvout.writerow(row)

import cando as cnd
ncpus=3
cnd.add_cmpds("train_smile.tsv", file_type='smi', fp='rd_ecfp4', vect="int", cmpd_dir=".", v="tox21_train")
cnd.generate_matrix(v="tox21_train", lib_path='.',
fp="rd_ecfp4", vect="int",
dist="dice", org='homo_sapien', bs='coach',
c_cutoff=0.0, p_cutoff=0.0, percentile_cutoff=0.0,
i_score="dCxP", out_file='homo_sapien-tox21_train.tsv', out_path="",
nr_ligs=True, approved_only=False,
lig_name=False, ncpus=ncpus)

But the following error is encountered. How to resolve this?

Error message:
Generating CANDO matrix...
Traceback (most recent call last):

File "", line 10, in
lig_name=False, ncpus=ncpus)

File "/home/mazharul/anaconda3/lib/python3.7/site-packages/cando/cando.py", line 4242, in generate_matrix
l_fps = pickle.load(f)

UnpicklingError: pickle data was truncated

Create 2 indication dictionaries (id to compound list and name to id) to replace indication object

Dear developer,
I just found out about this package, which is great, well, I am quite new to this. So, I am getting a little bit confused about input?
Can I input like some proteins or uniprot ID which in some certain pathway to predict which compound could activate/deactivate it? How can I do this? And, these output compounds are already in your database or can be selected from another big database like zinc? I am not sure if I make myself clear and hope you can understand. Thank you for your valuable time and sorry for these silly questions.