Developer builds pure Python DNA genome assembler using Branch and Bound algorithm

·1 views

A developer has built a DNA genome assembler from scratch in Python without using any external libraries, tackling the NP-hard Shortest Common Superstring (SCS) problem. The project models overlaps between DNA reads as a directed graph, reducing the SCS problem to the well-known Asymmetric Travelling Salesman Problem by introducing a dummy node. A Branch and Bound algorithm was implemented to find the optimal TSP path, with Kruskal's Minimum Spanning Tree algorithm serving as the lower bound for the search. The developer acknowledged that optimized libraries exist for genome assembly but chose to build everything manually to deepen understanding of NP-hard problem-solving. The project has been shared on GitHub, with the author seeking community feedback on performance bottlenecks and improved heuristics.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

AI-Generated Code Builds Hidden Technical Debt, Developer Warns

A developer working on a personal side project discovered he could not fix a simple bug because he had never truly understood the AI-generated code he had been using for weeks. He describes the experience as a 'loan' — AI writes the code, but the developer still owes the time needed to understand, debug, and maintain it. Research on AI-assisted commits has found higher rates of correctness and security issues compared to human-written code, suggesting the risks are not merely anecdotal. Developer surveys also show a correlation between heavy AI tool reliance and increased burnout, which the author says matched his own experience once the productivity boost faded. Rather than abandoning AI tools, he recommends treating generated code like a colleague's work — reading it carefully, section by section, until it is genuinely understood.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Claude Code left 10 CPU-spinning zombie processes running for nearly two days

A software developer discovered that a Claude Code AI session had left behind ten orphaned shell processes silently consuming over 700% CPU on his 10-core laptop for nearly two days. The processes were remnants of a deliberate load-testing script that had been designed to saturate all CPU cores while running an integration test suite. Once the test finished, the cleanup step never executed, leaving the busy-loop spinners reparented to PID 1 and running indefinitely. The developer identified them by checking load averages and inspecting full process argument lists, which revealed elapsed runtimes of almost 48 hours. The incident highlights two failure points: orphaned background jobs that survive their parent session, and AI-generated scripts that may not reliably clean up after themselves.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

HTML Forms Explained: Key Tags, Elements, and Attributes for Beginners

HTML forms serve as the primary tool for collecting user input on webpages, using a main container tag that holds elements like text fields, checkboxes, radio buttons, and drop-down lists. Each form element has a specific role, and tags such as label, fieldset, and legend help organize forms for better readability. Attributes like type, name, placeholder, required, and readonly control how form elements behave and how data is submitted. The name attribute is especially important for grouping radio buttons, allowing browsers to enforce single selection within a group. Some attributes, such as min and max, function differently depending on the element — restricting user input in range sliders but only displaying a measured value in the meter element.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Fintech Engineer Builds In-House LLM Gateway to Avoid Handing Out AWS Credentials

A fintech infrastructure engineer faced a common dilemma when non-technical teams requested access to AWS Bedrock: sharing IAM credentials or a single shared key both posed serious security and cost-tracking risks. Distributing individual AWS keys risks credentials spreading across .env files and CI variables, while a shared key makes it impossible to attribute usage or isolate a runaway process draining the monthly budget. Rather than routing sensitive prompts through third-party hosted gateways like LiteLLM or Portkey, the engineer chose to keep all traffic within the company's own network perimeter. The solution was a self-hosted LLM gateway that issues internal proxy tokens to teams, while the real AWS credentials remain exclusively within the gateway process itself. The design allows per-key model restrictions and usage tracking, giving finance teams an auditable record without exposing cloud credentials to any downstream service or user.

0 comments Read more at DEV Community